/usr/share/acl2-7.1/memoize-raw.lisp is in acl2-source 7.1-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 | ; ACL2 Version 7.1 -- A Computational Logic for Applicative Common Lisp
; Copyright (C) 2015, Regents of the University of Texas
; This version of ACL2 is a descendent of ACL2 Version 1.9, Copyright
; (C) 1997 Computational Logic, Inc. See the documentation topic NOTE-2-0.
; This program is free software; you can redistribute it and/or modify
; it under the terms of the LICENSE file distributed with ACL2.
; This program is distributed in the hope that it will be useful,
; but WITHOUT ANY WARRANTY; without even the implied warranty of
; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
; LICENSE for more details.
; Written by: Matt Kaufmann and J Strother Moore
; email: Kaufmann@cs.utexas.edu and Moore@cs.utexas.edu
; Department of Computer Science
; University of Texas at Austin
; Austin, TX 78712 U.S.A.
; memoize-raw.lisp -- Raw lisp definitions for memoization functions, only to
; be included in the hons-enabled ACL2 executables.
; The original version of this file was contributed by Bob Boyer and Warren
; A. Hunt, Jr. The design of this system of Hash CONS, function memoization,
; and fast association lists (applicative hash tables) was initially
; implemented by Boyer and Hunt. Contributions have been made since then by
; Jared Davis, Matt Kaufmann, J Moore, and Sol Swords.
(in-package "ACL2")
; SECTION: To do, near-term
; - Incorporate some Lisp comments into the :doc. In particular, probably the
; *record-xxx* and *report-xxx* should be incorporated into the :doc. In
; general, add/improve documentation, including for the following.
; - *memoize-summary-order-list*
; - *sort-to-from-by-calls*
; - memoize (e.g., clarify/emphasize that :recursive actually avoids inlining
; and :recursive nil avoids counting recursive calls in (memsum)); maybe
; explain how fns like LEN with special raw-Lisp code can thus be memoized
; only with :recursive nil; perhaps move some things from the Lisp comment
; to the :doc
; - Memoize-summary and memsum. Incorporate long comment in memoize-summary.
; Update :doc (maybe other topics too) to clarify that we stop at
; most-positive-mfixnum [see safe-incf]. Alternatively, memsum could just
; print a message if we ever hit that limit, and require an optional
; argument if one wants the results anyhow.
; - Update :doc for the pons/hons-wash soundness fix checked into git,
; especially for hons-wash.
; - Consider counting pons calls even when a number is returned; else make sure
; the :doc is clear about this. (See the definition of pons.)
; - Perhaps write an essay summarizing how memoization works, including
; discussion of *memoize-call-array* (or at least pointing to the comment
; there).
; - Add interface functions, perhaps including the following. I can imagine a
; single macro or function that takes, say, a keyword argument in order to
; print (to the comment window, or to state perhaps using the oracle) what's
; desired. OR -- maybe suitable utilities can go into books, if we make
; enough interface functions available.
; o *Memoize-summary-order-list* (as a state global perhaps)
; o Profile, unmemoize-profiled; maybe even profile-all and profile-acl2, but
; maybe keep those instead in books/centaur/memoize/old/profile*, or maybe
; both, with a sort of autoloading as is done for WET.
; o And so on -- see in particular function with this comment: "Note: This
; function should probably either be made available in the loop or else
; eliminated."
; - If we want to be able to memoize n functions, then it looks like the number
; of (virtual) columns of *memoize-call-array* need only be something like
; n+5. So perhaps that array need not be a square array -- it can have fewer
; columns than rows -- in which case we can save close to a factor of two on
; space. But that might well require considerable code re-work; for example,
; memoize-call-array-grow computes (/ *2max-memoize-fns* 2), but surely that
; would have to change. It's not clear that it's worth saving a factor of
; roughly two on something whose default is of size 10^6 and might not need
; to grow.
; - We have calls like (make-symbol "BODY-NAME") and bindings like
; (,*mf-old-caller* *caller*) -- the former interns a new symbol each time
; (though with the same, EQ result each time) while the latter saves a symbol
; using a defconstant form in file acl2.lisp. Consider making this all more
; consistent in style.
; SECTION: To do, longer-term
; - Consider eliminating the function memoized-values (and memstat) or, perhaps
; better yet, have it return a data structure rather than printing individual
; values.
; - Replace (COND (T <code>)) by <code> in what is generated by
; memoize-fn-def. (For an example, look just below memoize-fn-def.)
; - Perhaps think about this comment in the definition of pons:
; It might be tempting to think about creating a static cons here.
; - Consider using memo-max-sizes-entry records not only to guess good sizes,
; but also to guess good rehash sizes.
; - Tests such as (< (* 100 local-tt) tt) and (> (* 10000 local-tt) tt) in
; function memoize-summary-after-compute-calls-and-times restrict reporting
; to cases where the time is at least 1% of the time taken by the main
; function. We might want to make this customizable.
; - Consider finer-grained way to make memoization thread-safe.
; - Consider eliminating defun-one-output in most cases, simply for the sake of
; elegance.
; - Consider eliminating some uses of our-syntax, instead relying on normal
; ACL2 printing.
; - Look at the following comment in memoize-fn-outer-body and consider the
; possibility of using clrhash in some cases (e.g., there and in
; clear-one-memo-and-pons-hash) to save spec.
; Below, we "clear" the tables by setting them to nil, rather than by calling
; clrhash. The latter approach would probably avoid reallocation of space, but
; we suspect that such a gain in space efficiency might be offset by a loss in
; time efficiency. The present approach has been working, so we prefer not to
; change it. Below is just a little space analysis.
; - Consider printing sufficiently small floats without E notation.
; - Think about when to use defv vs. defg; maybe pick one and stick with it?
; Or perhaps better yet, this question might be moot after considering the
; next item.
; - Perhaps bundle up user-settable variables (see "User-settable variables for
; recording" below) into a single defrec stored in a state global, and
; provide suitable interface functions. Careful though -- some books may
; bind current such variables, so we'd probably want to provide a convenient
; mechanism for doing such bindings. As Jared notes, if such records are
; stored in ordinary special variables -- i.e., not introduced with defg or
; defv -- then although some performance advantage of defg/defv might be
; lost, it can be easier to support multi-threading by taking advantage of
; the fact that bindings of specials are thread-local. He also reminds us of
; the ability to use local variables [see for example mf-shorten and the use
; of physical-memory-cached-answer in the defun of physical-memory], and
; tells us that these can give similar efficiency to defv/defg with the
; similar drawback of not being thread-local, but with the advantage that the
; locality can ease not only ACL2 source code maintenance, but also avoid
; problems for other people's code (because it can't access such globals
; directly).
; - Document user-settable variables (or if the item just above is implemented
; with a record stored in a state global, document the fields and/or their
; readers and setters).
; - Try turning egc on in CCL. (See function acl2h-init.) Jared says (9/2014)
; that "leaving on EGC could really reduce the memory requirements of 95+% of
; the ACL2 jobs in the community books, and could be really good."
; - Consider moving start-sol-gc and supporting functions out of the ACL2 code
; base, quite possibly into community books directory
; books/centaur/misc/memory-mgmt/. But perhaps create some way (an interface
; of some sort) to get the effect of start-sol-gc, when desirable, without
; the need for trust tags. Perhaps consider the effect on ACL2(hp). Jared
; thinks that removing memory management responsibility out of the ACL2
; sources "might be especially useful if you are going to turn EGC back on by
; default."
; - In pons-addr-of-argument, (+ hl-dynamic-base-addr (hl-staticp x)) is
; probably a fixnum. Especially for GCL, maybe wrap (the fixnum ...) around
; it. Also maybe declaim a return type of (values (or null fixnum)), in case
; that helps GCL in particular; or maybe make this a macro (or at least
; inline it, but then maybe we still want the declaim form).
; - Revisit the comment in mf-index about how we use hl-staticp without
; checking that the argument is as normally expected.
; - Consider this suggestion from Jared, quoting a comment in memoize:
; Jared Davis suggests that we consider bundling up these 13 parameters, for
; example into an alist. He says: "Various subsets of these arguments occur in
; spaghetti fashion throughout the code for memoize, add-trip, the
; memoize-table stuff, etc."
; - Investigate odd behavior of with-lower-overhead for LispWorks, as explained
; in a comment there.
; SECTION: Preliminaries
;; One may comment out the following PUSHNEW and rebuild to get profiling
;; times not based upon the curious and wondrous RDTSC instruction of some x86
;; processors. On a machine with several cores, the RDTSC values returned by
;; different cores may be slightly different, which could lead one into such
;; nonsense as instructions apparently executing in negative time, when timing
;; starts on one core and finishes on another. To some extent we ignore such
;; RDTSC nonsense, but we still can report mysterious results since we have no
;; clue about which core we are running on in CCL.
#+ccl
(eval-when
(:execute :compile-toplevel :load-toplevel)
(when (fboundp 'ccl::rdtsc)
(pushnew :RDTSC *features*)))
; SECTION: Multithreaded Memoization
; This section was initially contributed by Jared Davis, who contributed the
; following comment (indented and re-filled here).
; The implementation is in many ways suboptimal because, for the most part, I
; just want to get something working. To keep things simple, I have added a
; single, global memoize lock. This lock needs to be acquired whenever a
; thread needs to do something that could mutate soundness-related memoize
; state. This is straightforward, but could (obviously) lead to a lot of
; lock contention if you write multithreaded code that is using memoization
; heavily. It could also slow down single threaded code because of the
; overhead of acquiring and releasing locks all the time.
;
; In the future it would probably be worth reworking things so that, e.g.,
; each memo table has its own separate lock. On the other hand, that makes
; things like deadlock a lot trickier. That sounds pretty horrifying, so I'm
; going to stick with the simple but possibly slow implementation for now.
(defg *enable-multithreaded-memoization*
;; By default set up multithreaded memoization only if we are running on
;; ACL2(p). However, nothing here is really ACL2(p) specific, and users of
;; plain ACL2(h) can change this in raw Lisp if they know what they're doing.
;;
;; Note for raw Lisp hackers. We ordinarily consult this variable
;; dynamically. However, to make memoized functions execute more
;; efficiently, we only consult its value AT MEMOIZE TIME when creating the
;; memoized definitions. Consequences of this:
;;
;; - If you ever want to change this, e.g., because you are using ordinary
;; ACL2(h) instead of ACL2(hp), but you want to do something with raw
;; Lisp threads, then you really need to do something like this:
;;
;; (setq *enable-multithreaded-memoization* t)
;; (rememoize-all)
;;
;; - You need to be careful to do the rememoize-all AFTER you have already
;; loaded any books that might contain memoized functions. For instance,
;; if you do the above and then include a book with a memoized F, then
;; the definition of F will not do appropriate locking until you again do
;; a rememoize-all.
#+acl2-par t
#-acl2-par nil)
; Historical Note: Many function names below are prefixed with ``mf-''. This
; was a convention followed by Bob and Warren in their original prototype of
; this code, and preserved (and possibly followed) by Jared. We have continued
; to follow the convention but we confess we don't know why ``mf-'' was used or
; what it stood for! Our best guess is ``memoize function.'' The prefix
; ``ma-'' stood for ``memoize call array.''
(defun mf-multiprocessing (flg &optional quiet)
(cond ((iff *enable-multithreaded-memoization* flg)
(when (not quiet)
(format *standard-output*
"~%; Note: Multiprocessing is already ~a.~%"
(if *enable-multithreaded-memoization* "on" "off"))))
(t
(unwind-protect ; make sure we finish!
(setq *enable-multithreaded-memoization* flg)
(rememoize-all))))
flg)
(defg *global-memoize-lock*
(make-lock "*GLOBAL-MEMOIZE-LOCK*"))
(defmacro with-global-memoize-lock (&rest forms)
; We consult *enable-multithreaded-memoization* at runtime; compare with macro
; with-global-memoize-lock-static.
#-(or ccl sb-thread lispworks) ; with-lock-raw is just progn
`(progn ,@forms)
#+(or ccl sb-thread lispworks)
; We tried the following alternative to the code below, in order to avoid
; duplicating code during macroexpansion. But for ACL2(h) (in its normal mode
; -- that is without *enable-multithreaded-memoization*, hence not ACL2(hp))
; this seemed to slow down the regression slightly. This macro is only used in
; this file so we're not too concerned about the compiled code duplication.
; (let ((body (gensym)))
; ;; Avoid duplication of ,forms to avoid code blowup
; `(flet ((,body () ,@forms))
; (if *enable-multithreaded-memoization*
; (with-lock-raw *global-memoize-lock* (,body))
; (,body))))
`(if *enable-multithreaded-memoization*
(with-lock-raw *global-memoize-lock* ,@forms)
(progn ,@forms)))
(defmacro with-global-memoize-lock-static (&rest forms)
;; Faster but less safe way to grab the lock, i.e., this version consults
;; *enable-multithreaded-memoization* at macroexpansion time instead of at
;; run time. This is nothing more than a progn in the single-threaded case.
;;
;; NOTE: ONLY USE THIS WHEN CREATING MEMOIZED DEFINITIONS. For everything
;; else, the safer version should be plenty fast.
#-(or ccl sb-thread lispworks) ; with-lock-raw is just progn
`(progn ,@forms)
#+(or ccl sb-thread lispworks)
(if *enable-multithreaded-memoization*
`(with-lock-raw *global-memoize-lock* . ,forms)
`(progn . ,forms)))
(defmacro mf-mht (&rest args)
; Call this macro with :shared :default unless we know that the call
; (mf-multiprocessing t) will blow away the hash table (and reconstitute it as
; a :shared table).
; We believe that at least one of :lock-free or :shared must be t in order to
; avoid "not owner of hash table" errors, and the following experiments support
; that belief.
; (defparameter *ht* (make-hash-table :lock-free nil :shared t))
; ; ok:
; (ccl:process-run-function "t1" (lambda () (setf (gethash 'my-key *ht*) 17)))
; (defparameter *ht* (make-hash-table :lock-free t :shared nil))
; ; ok:
; (ccl:process-run-function "t2" (lambda () (setf (gethash 'my-key *ht*) 17)))
; (defparameter *ht* (make-hash-table :lock-free nil :shared nil))
; ; error:
; (ccl:process-run-function "t3" (lambda () (setf (gethash 'my-key *ht*) 17)))
; The following experiment suggests that we get better performance with :shared
; t than with :lock-free t. In fact :shared t may rival performance with
; :shared nil.
; cd books/centaur/tutorial/
; (include-book "alu16-book")
; (time$ (def-gl-thm foo
; :hyp (and (alu16-test-vector-autohyps)
; (equal op *op-mult*)
; (< a (expt 2 12))
; (< b (expt 2 12)))
; :concl (equal (alu16-basic-result)
; (mod (* a b) (expt 2 24)))
; :g-bindings (alu16-default-bindings)))
; :lock-free nil :shared nil
; 48.42 seconds realtime, 48.29 seconds runtime
; (4,169,078,496 bytes allocated).
; :lock-free t :shared nil
; 54.33 seconds realtime, 54.19 seconds runtime
; (4,528,524,384 bytes allocated).
; :lock-free nil :shared t
; 48.35 seconds realtime, 48.23 seconds runtime
; (4,169,087,440 bytes allocated).
; To access or update tables created by mf-mht, we use mf-gethash and
; mf-sethash. Of course, :shared t is not sufficient to avoid the use of locks
; when performing atomic operations. Moreover, as of this writing only CCL,
; among the supported host Lisps, pays attention to :shared (or :lock-free).
; Pons tables contain flex-alists, which can contain hash-tables. Since
; hons-raw.lisp is loaded before memoize-raw.lisp, we define our own interface
; with hl-flex-acons; see mf-flex-acons. See "SECTION: Ponsing: Creating keys
; ("ponses") for memoization tables".
(if (assoc-keyword :shared args) ; then use :shared as supplied
`(hl-mht ,@args)
`(if *enable-multithreaded-memoization*
(hl-mht :shared :default ,@args)
(hl-mht ,@args))))
(defmacro mf-gethash (key ht)
; Ht was created by mf-mht. See mf-mht.
`(gethash ,key (the hash-table ,ht)))
(defmacro mf-sethash (key val ht)
; Ht was created by mf-mht. See mf-mht.
`(setf (gethash ,key (the hash-table ,ht)) ,val))
(defmacro mf-flex-acons (elem al)
`(hl-flex-acons ,elem ,al *enable-multithreaded-memoization*))
(defmacro mf-hash-table-count (ht)
; Ht was created by mf-mht. See mf-mht.
`(hash-table-count (the hash-table ,ht)))
(defmacro mf-hash-table-size (ht)
; Ht was created by mf-mht. See mf-mht.
`(hash-table-size (the hash-table ,ht)))
(defmacro mf-maphash (lam ht)
; Ht was created by mf-mht. See mf-mht.
`(maphash ,lam (the hash-table ,ht)))
; SECTION: Constants and globals
; This section contains most of the constants and globals in this file,
; including all globals that might be modified (for example, directly by the
; user, or during initialization).
; NOTE: It might be best to skip this section on a first reading, or at most
; browse it lightly, returning to look at specific constants and globals when
; they are encountered later.
; Subsection: User-settable variables for recording
; To minimize metering overhead costs, one may set the "*RECORD-" variables
; below to NIL before memoizing.
; *RECORD-BYTES* and other *RECORD-...* variables are bound in
; REMEMOIZE-ALL, so we use DEFPARAMETER rather than DEFG.
; Warning: Keep the following set of globals in sync with memoize-info, as
; described there.
(defparameter *record-bytes*
; If *record-bytes* is not NIL at the time a function is memoized, we attempt
; to count all heap bytes allocated during calls of that function.
; It is quite possibly fine to replace the following form by T. However, we
; leave this restriction, which produces nil in 32-bit CCL, because it has
; probably been here for many versions through 6.5 and at this point (August
; 2014) we don't expect much use of 32-bit CCL when doing memoization.
(and (or (member :ccl *features*) ; most Lisps don't report bytes
(member :sbcl *features*))
(> most-positive-fixnum (expt 2 32))))
(defparameter *record-calls*
; If *RECORD-CALLS* is not NIL at the time a function is memoized, we count all
; calls of the function.
t)
(defparameter *record-hits*
; If *RECORD-HITS* is not NIL at the time a function is memoized, we count the
; number of times that a previously computed answer is used again.
t)
(defparameter *record-mht-calls*
; If *RECORD-MHT-CALLS* is not NIL at the time a function is memoized, we
; record the number of times that a memoize hash-table is created for the
; function. See *report-mht-calls* below for an explanation of why we
; might create many hash-tables for a function.
t)
(defparameter *record-pons-calls*
; If *RECORD-PONS-CALLS* is not NIL at the time a function is memoized, pons
; calls are counted.
t)
(defparameter *record-time*
; If *RECORD-TIME* is not NIL the time a function is memoized, we record the
; elapsed time for each outermost call of the function.
t)
; Subsection: User-settable variables for reporting
(defg *mf-print-right-margin* ; rather arbitrary default; seems to work well
79)
(defv *report-bytes*
; When memoize-summary is invoked, print the number of bytes allocated on the
; heap, if this variable is not nil at that time and the host Lisp supports
; providing that information.
#+(or ccl sbcl) t #-(or ccl sbcl) nil)
(defv *report-calls*
; When memoize-summary is invoked, print the number of calls, if this variable
; is not nil at that time.
t)
(defv *report-calls-from*
; When memoize-summary is invoked, print the following if this variable is not
; nil at that time: which functions called the given function, how many times,
; and how long the calls took.
t)
(defv *report-calls-to*
; When memoize-summary is invoked, print the following if this variable is not
; nil at that time: which functions were called by the given function, how many
; times, and how long the calls took.
t)
(defv *report-hits*
; When memoize-summary is invoked, print the number of times that a previously
; computed answer was reused, if this variable is not nil at that time.
t)
(defv *report-mht-calls*
; When memoize-summary is invoked, print the number of times that a memo
; hash-table for the function was created, if this variable is not nil at that
; time. This may be of interest to those who memoize functions with stobj
; arguments, since when such a stobj changes, the function's memoization hash
; table is deleted, and then later, when necessary, created afresh.
t)
(defv *report-pons-calls*
; When memoize-summary is invoked, print the number of calls of pons, if this
; variable is not nil at that time.
t)
(defv *report-time*
; When memoize-summary is invoked, print information about the total time used
; to compute the outermost calls, if this variable is not nil at that time.
t)
(defv *report-on-memo-tables*
; When memoize-summary is invoked, print information about memo tables, if this
; variable is not nil at that time.
t)
(defv *report-on-pons-tables*
; When memoize-summary is invoked, print information about pons tables, if this
; variable is not nil at that time.
t)
(defg *sort-to-from-by-calls* nil)
(defg *memoize-summary-order-reversed*
; When memoize-summary is invoked, report on functions in order from least to
; greatest, if this variable is not nil at that time.
nil)
(defg *mf-print-alist-width* 45)
(defg *memoize-summary-order-list*
; This is a list of functions that MEMOIZE-SUMMARY uses to sort all functions
; that are currently memoized in preparation for displaying information about
; them. The order used is lexicographical, with the first order having the
; most weight. Each order function takes one argument, a symbol, and returns a
; rational.
; The default is '(total-time number-of-calls).
; Options for the functions include the following.
; bytes-allocated [CCL and SBCL only]
; bytes-allocated/call [CCL and SBCL only]
; execution-order
; hits/calls
; pons-calls
; number-of-calls
; number-of-hits
; number-of-memoized-entries
; number-of-mht-calls
; symbol-name-order
; time-for-non-hits/call
; time/call
; total-time
'(total-time number-of-calls))
(defg *memoize-summary-limit*
; The value is either nil or the maximum number of functions upon which
; memoize-summary reports. A nil value means report on all."
100)
(defg *condition-nil-as-hit*
; Set this variable to a non-nil value if you want the failure of a memoization
; condition to be recorded as a "hit". That seems counterintuitive, which is
; why the default is nil. (Note profiling doesn't report misses, so this
; decision doesn't impact profiling reports.) But if you care a lot about the
; "Time per missed call" reported by memsum, maybe you'll want to set this to
; t. NOTE that this value is consulted at the time a function is memoized, not
; at call time or memsum time.
nil)
; Subsection: Miscellaneous user-settable variables
(defvar *never-memoize-ht*
; Function exit-boot-strap-mode further populates this hash table by calling
; function initialize-never-memoize-ht.
; Users can set this with never-memoize.
; Note: Consider locking uses of this global variable.
(let ((ht (mf-mht :test 'eq :shared :default)))
(loop for x in *stobjs-out-invalid*
do (mf-sethash x t ht))
#+ccl (mf-sethash 'ccl::rdtsc t ht) ; used in memoize implementation
ht))
; Subsection: Variables not user-settable (managed by the implementation)
(defg *float-ticks/second* (float (expt 2 31))) ; 2 GHz
(defg *float-ticks/second-initialized* nil)
(declaim (float *float-ticks/second*))
(defg *pons-call-counter* 0)
(defg *pons-misses-counter* 0)
(declaim (type mfixnum *pons-call-counter*))
(declaim (type mfixnum *pons-misses-counter*))
(defg *mf-shorten-ht*
; This global is used in mf-shorten. As explained there, no lock is needed.
; Since hons-copy is applied to form the keys, we can use eql as the test.
(mf-mht :test 'eql :shared :default))
(defun initial-memoize-info-ht ()
(if *enable-multithreaded-memoization*
(mf-mht :shared :default)
; This table is essentially blown away and reconstituted by rememoize-all,
; hence by (mf-multiprocessing t). So we avoid using :shared :default below,
; in order to make the table a bit more efficient in the uniprocessing case --
; though probably efficiency isn't particularly important for this hash table.
(mf-mht)))
(defg *memoize-info-ht*
; This hash table associates memoized function symbols with various
; information. It has two mappings in one (see setf forms in memoize-fn), that
; is, two kinds of keys: symbols and numbers.
;
; 1. It maps each currently memoized function symbol, fn, to a DEFREC record
; of type MEMO-INFO-HT-ENTRY. One field of that record is :NUM.
;
; 2. It maps each such :NUM value back to the corresponding symbol, fn.
(initial-memoize-info-ht))
(declaim (hash-table *memoize-info-ht*))
(defparameter *memo-max-sizes*
; This hash table binds function names to memo-max-sizes-entry structures.
; Jared originally added this table because he wanted to know how big
; memoization tables were getting (so that he could set up appropriate initial
; sizes), but when tables are cleared they are thrown away, so for tables that
; are frequently cleared it wasn't possible to see how large the table had
; become.
; After seeing the information, it seemed a good idea to use it to infer what a
; good table size might be when the memo table is re-created.
(mf-mht :shared :default))
(defun initial-memoize-call-array (n)
; Warning: It is assumed in sync-memoize-call-array that this array has a
; length of 1 when n is 1.
(make-array n :element-type 'mfixnum :initial-element 0))
(defg *memoize-call-array*
(initial-memoize-call-array 1))
; Essay on the *Memoize-Call-Array*
; *Memoize-call-array*, `ma' for short, is a 1D array that is used to store
; information about how many times one memoized function calls another and how
; much time is spent in such calls. It also holds performance data on the
; callers themselves (independent of which subroutines they're calling).
; This array is conceptually two-dimensional. To read from or write to this
; array, see ma-index and the ma-index-xxx macros. Each "dimension" of ma is
; twice the maximum number of memoized functions; so the length is 4 times the
; square of the maximum number of memoized functions. This is a very
; inefficient scheme; roughly half of the allocated space is unused! See
; below.
; The length of this array is both a fixnum and an mfixnum; see
; memoize-call-array-grow.
; Ma is initialized in memoize-init. Think of ma as a two dimensional array
; with dimensions (twice the max number of memoized functions) x (twice the max
; number of memoized functions).
; COLUMNS. Each 'column' corresponds to info about a memoized function, but
; the first five columns are 'special'. We count rows and columns starting at
; 0. Column 0 is used as scratch space by compute-calls-and-times for sums
; across all functions. Columns 1, 2, and 3 are not currently used at all.
; Column 4 is for the anonymous 'outside caller'. Column 5 is for the first
; memoized function, column 6 is for the second, and so on.
; ROWS. In columns 5 and greater, row 0 is used to count 'bytes', row 1 counts
; 'hits', row 2 counts mht calls, and row 4 counts pons calls that don't simply
; return a number (i.e., an address in the static-hons case). (3 was formerly
; used to count hons calls, but that is no longer the case.) The elements of
; an ma column starting at row 10 (twice the number assigned to the first
; memoized function) are for counting and timing info. Suppose column 7
; corresponds to the memoized function FOO and column 12 corresponds to the
; memoized function BAR. Whenever FOO calls BAR, element 2*12 of column 7 will
; be incremented by 1, and the total elapsed ticks for the call will be added
; to element 2*12+1 of column 7. See the picture below.
; Though ma may 'grow', it may not grow while any memoized function is running,
; and here is why: every memoized function has a cached opinion about the size
; of ma. (Here is a technical explanation, which one may wish to ignore. For
; example, memoize-fn binds the variable fn-col-base to (ma-index fnn), which
; is computed based on the size of ma at the time the function is memoized.
; Fn-col-base is passed to memoize-fn-outer-body, where it is used to generated
; code for the memoized function.)
; Here is an ASCII picture of the *memoize-call-array* with a key to explain
; what is stored where.
; <--------------max---------------><--------------max------------>
; 0 1 2 3 4 5 i ^
; |
; 0 [ ][x][x][x][ ][ ] ... [b] ... [x][x][x][x] ... [x][x][x][x] |
; 1 [ ][x][x][x][ ][ ] ... [h] ... [x][x][x][x] ... [x][x][x][x] |
; 2 [ ][x][x][x][ ][ ] ... [m] ... [x][x][x][x] ... [x][x][x][x] |
; 3 [ ][x][x][x][ ][ ] ... [x] ... [x][x][x][x] ... [x][x][x][x] |
; 4 [ ][x][x][x][ ][ ] ... [p] ...
; 5 [ ][x][x][x][ ][ ] ... [ ] ... ... ... max
; ... ... ...
; 2j [s][x][x][x][ ][ ] ... [c] ... [x][x][x][x] ... [x][x][x][x] |
; 2j+1 [s][x][x][x][ ][ ] ... [t] ... [x][x][x][x] ... [x][x][x][x] |
; ... ... ... |
; [ ][x][x][x][ ][ ] ... ... [x][x][x][x] ... [x][x][x][x] |
; [ ][x][x][x][ ][ ] ... ... [x][x][x][x] ... [x][x][x][x] v
; ... ... ... [x][x][x][x] ... [x][x][x][x] ^
; ... ... ... [x][x][x][x] ... [x][x][x][x] |
; ... ... ... ... ... max
; ... ... ... [x][x][x][x] ... [x][x][x][x] |
; [ ][x][x][x][ ][ ] ... ... [x][x][x][x] ... [x][x][x][x] |
; [ ][x][x][x][ ][ ] ... ... [x][x][x][x] ... [x][x][x][x] v
; Dimensions: (2 * max number of memoized fns)^2
; Contents:
; s -- scratch space for compute-calls-and-times as it sums across all functions
; (for calls in even-numbered rows and ticks in odd-numbered rows)
; b -- number of bytes used in calls of fni
; h -- number of hits on calls of fni
; m -- number of mht calls on fni
; x -- unused
; p -- number of pons calls for fni
; c -- number of times fni called fnj
; t -- number of ticks spent when fni called fnj
; Each memoized function symbol is assigned a number. The first ``real''
; function number is 5. Because rows 0-4 of every function's column are
; devoted to performance data about that function, and due to a poorly realized
; desire to make a function's number serve both as its column number and its
; row number, we end up wasting columns 0-4 in every row because we use rows
; 0-4 in every column (!). We don't entirely waste them: column 0 is used for
; scratch space and column 4 is used for the generic outside caller, but we
; have not found a use for columns 1, 2, or 3, or for column 0 of the odd rows.
; The uniformity of treating a function's number both as its column and row
; number is ``poorly realized'' because two rows are allocated to every
; function, e.g., data for function j resides in row 2j and row 2j+1, so the
; actual row number is not the function's number after all!
; Clearly this whole scheme -- evolved over several steps from Boyer and Hunt's
; original prototype -- is very wasteful but we are reluctant to change it
; simply because it works and we have not apparently suffered for lack of space
; for ma.
; The functions ma-index-calls and ma-index-ticks return index in the actual 1D
; array containing the call count and tick count, respectively, of the
; corresponding call from function i to function j.
; The following function, which is not part of our code, illustrates how to map
; from a pair of symbols, parent and child (caller and callee), to the index of
; the *memoize-call-array* containing the number of times parent called child.
; It calls symbol-to-fixnum, which maps a memoized function symbol to its
; natural number counterpart.
; (defun call-count (parent child)
; ; Return the number of times parent has called child, as per
; ; *memoize-call-array*.
; (let* ((parent-index (symbol-to-fixnum parent))
; (child-index (symbol-to-fixnum child))
; (calls-index (ma-index-calls child-index parent-index))
; (calls (aref *memoize-call-array* calls-index)))
; (values calls
; :parent-index parent-index
; :child-index child-index
; :calls-index calls-index)))
; For example,
; (defun f1 (x) (declare (xargs :guard t)) (cons x x))
; (defun f2 (x) (declare (xargs :guard t)) (cons (f1 x) (f1 x)))
; (memoize 'f1)
; (memoize 'f2)
; (f2 7)
; (f2 8)
; (f2 9)
; :q
; ; These return T:
; (equal (call-count 'f2 'f1) 6)
; (equal (call-count 'f1 'f2) 0)
; ; This gives more complete information, namely call-count,
; ; parent index, child index, index of call count:
; (call-count 'f2 'f1)
; The assignment of numbers to function symbols is handled by
; *memoize-info-ht*. Use (maphash (function (lambda (fn val) (print (list fn
; val)))) *memoize-info-ht*) to inspect.
; As of this writing (Version_7.0), the number 4 is assigned to the
; non-function "outside-caller" (denoting the caller of outermost call of a
; memoized function) and 5-7 are assigned to the first three built-in functions
; listed below; they are always memoized by explicit calls to memoize in
; boot-strap-pass-2. Of course, this set and the particular numbers may change
; in future releases, but we include this information here to indicate aspects
; of the basic scheme. BAD-LISP-OBJECTP may not always be memoized, i.e., it
; is not in ACL2(hp). See *thread-unsafe-builtin-memoizations* for a list of
; functions that are not memoized in ACL2(hp).
; fn number assigned
; "outside-caller" 4
; FCHECKSUM-OBJ 5
; EXPANSION-ALIST-PKG-NAMES-MEMOIZE 6
; WORSE-THAN-BUILTIN 7
; BAD-LISP-OBJECTP 8
; By the way, the assignment of a unique number to each memoized function is
; done initially by symbol-to-fixnum-create.
(declaim (type (simple-array mfixnum (*))
*memoize-call-array*))
(defg *callers-array*
; At the end of a call of compute-calls-and-times, which is called by
; memoize-summary, (aref *callers-array* n) will contain the numbers of the
; functions that have called the function numbered n.
(make-array 0))
(declaim (type (simple-array t (*)) *callers-array*))
(defconstant *ma-bytes-index* 0)
(defconstant *ma-hits-index* 1)
(defconstant *ma-mht-index* 2)
; Originally there was an *ma-hons-index* of 3, but we no longer count honses.
(defconstant *ma-pons-index* 4)
; We associate memoized function symbols with fixnums, and vice versa. See
; symbol-to-fixnum-create.
(declaim (type (and mfixnum
; Normally we are happy to declare variables in this file to be of mfixnum type
; only, but here we also declare these of type fixnum so that indices i below
; such variables may safely appear in forms (loop for i fixnum ...).
fixnum)
*max-symbol-to-fixnum*
*initial-2max-memoize-fns*
; *initial-max-symbol-to-fixnum* ; defconstant, below
*2max-memoize-fns*))
(defconstant *initial-max-symbol-to-fixnum* 4)
(defg *max-symbol-to-fixnum* *initial-max-symbol-to-fixnum*)
(defv *initial-2max-memoize-fns* 1000)
(defg *2max-memoize-fns* ; length of *memoize-call-array*
*initial-2max-memoize-fns*)
(defg *max-mem-usage*
; This global is set in start-sol-gc. It is an upper bound, in bytes of memory
; used, that when exceeded results in certain garbage collection actions.
; See also the centaur/misc/memory-mgmt books.
(expt 2 32))
(defg *gc-min-threshold*
; This is set in start-sol-gc.
; See also the centaur/misc/memory-mgmt books.
(expt 2 30))
(defg *memoize-init-done* nil)
(defg *hons-gentemp-counter* 0)
(defvar *sol-gc-installed* nil)
(defg *memoize-use-attachment-warning-p* t)
(defv *number-of-arguments-and-values-ht*
; Mf-note-arity is an interface for updating this variable.
; This hash table maps a symbol fn to a cons pair (a . d), where a is the
; number of inputs and d is the number of outputs of fn. NIL for a or d
; indicates 'don't know'.
(mf-mht :shared :default))
(declaim (hash-table *number-of-arguments-and-values-ht*))
; Globals not included in this section:
; - *caller*
; - *10-days*
; - *float-internal-time-units-per-second*
; SECTION: Our-syntax
(defmacro our-syntax (&rest args)
; This macro extends Common Lisp's with-standard-io-syntax by binding *package*
; and *readtable*. It may not be necessary to bind *readtable* here, but it
; has been done historically and it seems harmless. We bind *package* because
; the Common Lisp utility with-standard-io-syntax binds *package* to the
; CL-USER package.
; The settings in our-syntax are oriented towards reliable, standard, vanilla,
; mechanical reading and printing, and less towards human readability.
; Please, before changing the following, consider existing uses of this macro
; insofar as the changes might impact reliable, standard, vanilla, mechanical
; printing.
; Consider using our-syntax-nice.
`(with-standard-io-syntax
(let ((*package* (find-package-fast
(f-get-global 'current-package *the-live-state*)))
(*readtable* *acl2-readtable*))
,@args)))
(defmacro our-syntax-nice (&rest args)
; Our-syntax-nice may offer slightly more pleasant human readabilty than
; our-syntax. Although we stopped using it ourselves in Sept. 2014, we leave
; it for now since it is used in some books.
`(our-syntax
(setq *print-case* :downcase)
(setq *print-pretty* t)
(setq *print-readably* nil)
(setq *print-right-margin* ,*mf-print-right-margin*)
(setq *print-miser-width* 100)
,@args))
; SECTION: Number of Arguments
(defun mf-note-arity (fn nargs nvals)
; This raw Lisp function declares that fn has the indicated number of inputs
; and outputs.
; Note: Consider locking uses of this function, since it updates a global
; hash-table.
(mf-sethash fn (cons nargs nvals) *number-of-arguments-and-values-ht*))
(defun memoize-look-up-def-raw (sym quiet-p)
; This function attempts to return the definition of sym as an s-expression
; (not a symbol-function), as (name formals . rest). When that attempt fails,
; then an error occurs if quiet-p is nil but nil is returned otherwise.
; For CCL, we could evaluate the following two forms so that
; function-lambda-expression will return the form used in the defun of
; a function symbol.
; (setq ccl::*save-definitions* t)
; (setq ccl::*fasl-save-definitions* t)
; However, while memoization is implemented with some features specific to CCL
; (e.g., RDTSC), in general we prefer that the functionality be independent of
; the host Lisp, so that books are less likely to fail certification when
; changing host Lisps (or versions of a given host Lisp). Moreover, an ACL2
; regression based on CCL took close to 2% longer when including those forms in
; acl2.lisp:
; With the above forms:
; 33606.972u 498.459s 1:16:57.09 738.6% 0+0k 0+2676744io 0pf+0w
; Without the above forms:
; 32964.076u 475.053s 1:15:31.71 737.8% 0+0k 135096+2542696io 78pf+0w
; (- 1 (/ (+ 32964.076 475.053) (+ 33606.972 498.459)))
; = 0.019536536570964436
; (- 1 (/ (+ 3600 (* 60 15) 31.71) (+ 3600 (* 60 16) 57.09)))
; = 0.01849216714424018
; The property acl2-saved-def, which appears below, is installed when save-def
; is used; see for example bad-lisp-objectp.
(let ((temp (get sym 'acl2-saved-def)))
(cond (temp (cdr temp))
(t (let* ((fn (symbol-function sym))
(lam (and fn
#+gcl ; accommodate early GCL 2.6.13
(and (fboundp 'function-lambda-expression)
(funcall 'function-lambda-expression fn))
#-gcl
(function-lambda-expression fn))))
(cond (lam (assert (eq (car lam) 'lambda))
lam)
(quiet-p nil)
(t (error "Lisp cannot find a definition for ~
~s.~%Consider using ~s or supplying to ~
~%memoize(-fn) an explicit :CL-DEFUN ~
argument.~a"
sym
'save-def
#-ccl
""
#+ccl
(format
nil
"~%Alternatively, you might solve this ~%~
problem, for CCL only, by setting Lisp ~%~
globals CCL::*SAVE-DEFINITIONS* and ~%~
CCL::*FASL-SAVE-DEFINITIONS* to T before~%~
defining ~a (though this might slow down~%~
performance)."
sym)))))))))
(defun-one-output mf-len-inputs (fn)
; This function returns the length of the inputs of fn, or nil for "don't
; know".
(assert (symbolp fn))
(assert (fboundp fn))
(assert (not (macro-function fn)))
(assert (not (special-operator-p fn)))
(or (let ((pair (mf-gethash fn *number-of-arguments-and-values-ht*)))
(and (consp pair)
(integerp (car pair))
(car pair)))
(let ((formals
(getprop fn 'formals t 'current-acl2-world (w *the-live-state*))))
(and (not (eq t formals))
(length formals)))
(let ((def (memoize-look-up-def-raw fn t)))
(and def (length (cadr def))))))
(defun-one-output mf-len-outputs (fn)
; This function returns the length of the outputs of fn, or nil for "don't
; know".
(assert (symbolp fn))
(assert (fboundp fn))
(assert (not (macro-function fn)))
(assert (not (special-operator-p fn)))
(or (let ((pair (mf-gethash fn *number-of-arguments-and-values-ht*)))
(and (consp pair)
(integerp (cdr pair))
(cdr pair)))
(let* ((state *the-live-state*)
(w (w state)))
(and (not (eq t (getprop fn 'formals t 'current-acl2-world w)))
(len (stobjs-out fn w))))
(and (get fn 'acl2-saved-def)
; Save-def guarantees that there is a single output of the defined function.
1)))
; SECTION: Timing Utilities
; WARNING: Whenever using *float-ticks/second*, consider whether it should be
; properly initialized if that has not already been done. Here is the story:
; In order to set *float-ticks/second* to a reasonably accurate value, there is
; a one-time delay due to a call of sleep in float-ticks/second-init, which
; might be noticeable when starting up ACL2. We avoid this delay until we need
; the accuracy. For example, fix-ticks is used in the wrappers of memoized
; functions, making it performance-critical, but it only uses
; float-ticks/second to compute a heuristic sanity check; so we avoid that
; delay in fix-ticks. On the other hand, for many purposes we want
; *float-ticks/second* to be accurate, such as in calls of total-time (for
; printing memoize-summary information). For those, we tolerate the delay.
; The community book books/centaur/memoize/timer.lsp provides some potentially
; useful ideas, but as of this writing (late August 2014) we stick pretty close
; to what has previously been done here.
(defconstant *10-days-in-seconds* (* 10 24 60 60))
(defun 10-days ()
(let ((n (* *10-days-in-seconds*
(ceiling *float-ticks/second*))))
(when (> n most-positive-mfixnum)
; It's OK to comment out the following error and only do the setq. But we
; have the error here in order to catch confusion on our part. In one
; experiment circa Sept 2014 on a 2.66 GHz Mac we got this:
; ? (10-days)
; 1855425871872000
; ? most-positive-mfixnum
; 1152921504606846975
; ?
(error "~s is too large: ~s. This is a surprise!~%Please ~
contact the ACL2 implementors."
'(10-days)
n)
(setq n most-positive-mfixnum))
n))
(defg *10-days*
; We get a reasonable approximation based on *float-ticks/second*, which is
; improved after calling float-ticks/second-init.
(10-days))
(declaim (type mfixnum *10-days*))
(defg *float-internal-time-units-per-second*
(float internal-time-units-per-second))
(declaim (float *float-internal-time-units-per-second*))
(defmacro internal-real-ticks ()
; We use logand below instead of mod because it is faster (see below). This
; use of logand relies on the fact that the binary representation of
; most-positive-mfixnum is a sequence of ones.
; The following tests use values of i that are fixnums, because that is what we
; expect to encounter in most cases and they avoid the overhead of building
; bignums. But there is no reason to expect the overwhelming advantage of
; avoiding mod to disappear if bignums are involved. Indeed, if "to
; most-positive-mfixnum" is replaced by "(+ most-positive-mfixnum n)" below,
; convincing results are still obtained.
; (defun my-test-0 (n)
; (loop for i from (- most-positive-mfixnum n) to most-positive-mfixnum
; when (eql (the integer i)
; -1)
; do (error "Found 0!")))
;
; (defun my-test-1 (n)
; (loop for i from (- most-positive-mfixnum n) to most-positive-mfixnum
; when (eql (the integer (logand i most-positive-mfixnum))
; -1)
; do (error "Found 0!")))
;
; (defun my-test-2 (n)
; (loop for i from (- most-positive-mfixnum n) to most-positive-mfixnum
; when (eql (the integer (mod i most-positive-mfixnum))
; -1)
; do (error "Found 0!")))
;
; (time$ (my-test-0 10000000))
; (time$ (my-test-1 10000000))
; (time$ (my-test-2 10000000)) ; much slower than the others
; Note that (ccl::rdtsc) returns a fixnum; see
; http://trac.clozure.com/ccl/wiki/ReleaseNotes.
#+(and RDTSC (not 32-bit-target)) ; faster for 64
'(the-mfixnum (ccl::rdtsc))
#+(and RDTSC 32-bit-target) ; slower for 32
'(the-mfixnum (logand (ccl::rdtsc64) ; don't truncate at 32-bit CCL fixnum
most-positive-mfixnum))
#-RDTSC
'(the-mfixnum (logand (the-mfixnum (get-internal-real-time))
most-positive-mfixnum)))
(defun-one-output float-ticks/second-init ()
; [Jared] multithreading note: I think it's fine not to protect this with a
; lock, if two threads happen to recompute it, that's seems harmless enough.
(unless *float-ticks/second-initialized*
#+RDTSC (let ((i1 (ccl::rdtsc64))
(i2 (progn (sleep .1) (ccl::rdtsc64))))
(cond
((> i2 i1)
(setq *float-ticks/second*
(* 10 (float (- i2 i1)))))
(t ; retain initial value of *float-ticks/second*
(format t
"~%***WARNING***: Float-ticks/second-init ~
failed;~%using arbitrary default value.~%Memoize ~
timings may be unreliable.~%")
(force-output *standard-output*))))
#-RDTSC (setq *float-ticks/second*
*float-internal-time-units-per-second*)
(setq *10-days* (10-days))
(setq *float-ticks/second-initialized* t)
(check-type *float-ticks/second*
(and float (satisfies plusp)))))
; SECTION: Safe-incf
(defun safe-incf-aux-error (x inc where)
(declare (type mfixnum x inc))
(cond
(where
(error "~%; SAFE-INCF-AUX: ** Error: ~a."
(list :x x :inc inc :where where)))
(t (setf (the mfixnum x)
most-positive-mfixnum)))
nil)
(defmacro safe-incf-aux (x inc &optional where)
; This is essentially safe-incf, but where inc is known to have a positive
; integer value.
(cond
((not (or (symbolp inc)
(and (< inc most-positive-mfixnum)
(> inc 0))))
(safe-incf-aux-error x inc where))
((and (true-listp x) ; e.g., x is (aref ar (.. expr ..))
(equal (len x) 3)
(member (car x) '(aref svref))
(symbolp (nth 1 x))
(consp (nth 2 x)))
(let ((idx (make-symbol "IDX")))
`(let ((,idx (the-mfixnum ,(nth 2 x))))
(declare (type mfixnum ,idx))
(safe-incf (,(nth 0 x)
,(nth 1 x)
,idx)
,inc
',where))))
(t (let ((v (make-symbol "V")))
`(let ((,v (the-mfixnum ,x)))
(declare (type mfixnum ,v))
(cond ((<= ,v (the-mfixnum
(- most-positive-mfixnum
(the-mfixnum ,inc))))
(setf (the mfixnum ,x)
(the-mfixnum (+ ,v (the-mfixnum ,inc))))
nil)
(t (safe-incf-aux-error ',x ',inc ',where))))))))
(defmacro safe-incf (x inc &optional where)
; Summary: we increment nonnegative mfixnum x by nonnegative mfixnum inc,
; returning nil. But if the result is too large to be of type mfixnum, then we
; cause an error if where is non-nil, else we set x to most-positive-mfixnum.
; SAFE-INCF is a raw Lisp macro that behaves the same as INCF when both X and
; INC are nonnegative MFIXNUMs and their sum is a nonnegative MFIXNUM. In a
; call of (SAFE-INCF x inc), X must be a place that holds an MFIXNUM. INC must
; evaluate to an MFIXNUM. Both X and INC must evaluate without side effects,
; so that it is impossible to tell which was executed first or whether only one
; or both were executed. If INC is not positive, no update takes place at all.
; Otherwise, if the sum of the values of X and INC is not an MFIXNUM, which is
; tested without causing an error, a run-time error will be caused. Else, if
; the sum is an MFIXNUM then, as with INCF, the place X will be set to hold the
; sum of the old value of that place and the value of INC. The value returned
; by SAFE-INCF is NIL. Caution: INC may be evaluated first, which is why side
; effects are prohibited.
; An optional third parameter is merely to help with error location
; identification.
; In (SAFE-INCF (AREF A (FOO)) INC), (FOO) is only evaluted once.
; Same for SVREF.
; The argument, where, is only used in reporting errors. When where is
; non-nil, an error occurs if (+ x inc) is too large to be an mfixnum;
; otherwise, x is set to most-positive-mfixnum. Note that nil is always
; returned; the only purpose is to increment x, not return a value.
(cond ((integerp inc)
(if (<= inc 0)
nil
`(safe-incf-aux ,x ,inc ,where)))
((symbolp inc)
`(if (>= 0 (the-mfixnum ,inc))
nil
(safe-incf-aux ,x ,inc ,where)))
(t (let ((incv (make-symbol "INCV")))
`(let ((,incv (the-mfixnum ,inc)))
(declare (type mfixnum ,incv))
(if (>= 0 ,incv)
nil
(safe-incf-aux ,x ,incv ,where)))))))
; SECTION: Ponsing: Creating keys ("ponses") for memoization tables
; Pons differs from hons in that it does not honsify its arguments and in that
; it takes a hash table as a third argument. We use pons in memoization.
; We use pons instead of hons in memoization because we could not afford to
; honsify (using fast-alist-fork!, say) certain alists in certain biology
; tests. About the same time, we (gratuitously) decided to stop hons'ifying
; the output of memoized functions.
; Most of the comments and code in this section were initially derived from
; books/centaur/memoize/pons.lsp, authored by Jared Davis and, as stated there,
; "a descendant of the memoization scheme developed by Bob Boyer and Warren
; A. Hunt, Jr. which was incorporated into the HONS version of ACL2, sometimes
; called ACL2(h)."
; Pons is the critical function for generating memoization keys. To a rough
; approximation, here is how we memoize (F arg1 arg2 ... argN):
;
; PONS := (PIST* arg1 ... argN)
; LOOK := MemoTableForF[PONS]
; if (LOOK exists)
; return LOOK
; else
; RESULT := (F arg1 arg2 ... argN)
; MemoTableForF[PONS] = RESULT
; return RESULT
;
; In other words, we use (PIST* arg1 ... argN) to create the hash key for the
; arguments arg1 ... argN. And PIST* is defined in terms of PONS.
;
; (PIST*) = NIL
; (PIST* X1) = X1
; (PIST* X1 X2) = (PONS X1 X2)
; (PIST* X1 X2 X3) = (PONS X1 (PONS X2 X3))
; ... ...
;
; As its name suggests, PONS is similar to a HONS. The main difference is that
; whereas (HONS X Y) requires us to recursively generate a "canonical" version
; of X and Y, (PONS X Y) does not descend into its arguments.
;
; It is worth noting that in the 0 and 1-ary cases of PIST*, no PONSing is
; necessary! Because of this it can be considerably cheaper to memoize unary
; and zero-ary functions than higher-arity functions. Note also that as the
; arity of the function increases, the amount of ponsing (and hence its cost)
; increases.
;
; The soundness requirement on PIST* is that two argument lists should produce
; EQL keys only if they are pairwise EQUAL. This follows from a stronger
; property of PONS:
;
; (EQL (PONS A B) (PONS C D)) ---> (EQL A C) && (EQL B D).
;
; Why is this property sufficient? The 0-2 argument cases are trivial. Here
; is a sketch of the 3-ary case, which generalizes easily to the N-ary case:
;
; Our goal is to ensure:
;
; If (EQL (PIST* A1 B1 C1) (PIST* A2 B2 C2))
; Then (EQUAL A1 A2) && (EQUAL B1 B2) && (EQUAL C2 C2)
;
; Assume the hypothesis:
;
; (EQL (PONS A1 (PONS B1 C1)) (PONS A2 (PONS B2 C2)))
;
; Then from our PONS property it follows immediately that
;
; 1. (EQL A1 A2), and hence (EQUAL A1 A2) since EQL implies EQUAL.
; 2. (EQL (PONS B1 C1) (PONS B2 C2)).
;
; From 2, again from our PONS property it follows that
;
; 1. (EQL B1 B2), and hence (EQUAL B1 B2) since EQL implies EQUAL.
; 2. (EQL C1 C2), and hence (EQUAL C1 C2) since EQL implies EQUAL.
;
; Which is what we wanted to show.
;
; For our memoization scheme to be effective, it is desirable for PIST* to
; produce EQL keys when given pairwise-EQL argument lists. This follows easily
; if our PONS property holds in both directions, that is:
;
; (EQL (PONS A B) (PONS C D)) <---> (EQL A C) && (EQL B D).
;
; Okay, so how does PONS actually work? First we will introduce a "slow"
; ponsing scheme, and then explain an optimization that avoids slow ponsing in
; many cases.
;
;
; Slow Ponsing.
;
; The above discussion hides the fact that PONS and PIST* take an additional
; argument, called the Pons Table. This table is essentially a scheme for
; remembering which keys we have produced for argument lists that have been
; encountered thus far. Note that the act of PONSing implicitly modifies the
; Pons Table.
;
; The Pons Table is similar to the CDR-HT-EQL in the Classic Honsing scheme;
; see the Essay on Classic Honsing. That is, it is an EQL hash table that
; binds each
;
; Y -> { key : key is the key for (PONS X Y) }
;
; As in classic honsing, these sets are represented as Flex Alists. The basic
; implementation of (PONS X Y), then, is as follows:
;
; Y_KEYS := PonsTable[Y]
; XY_KEY := FlexAssoc(X, Y_KEYS)
; If (XY_KEY was found)
; return XY_KEY
; Else
; NewKey = (CONS X Y)
; Y_KEYS := FlexAcons(NewKey, Y_KEYS)
; PonsTable[Y] := Y_KEYS
; return NewKey
;
; In other words, we build a new (X . Y) cons and use it as the key, unless we
; had previously seen these same arguments and such a key is already available.
;
;
; Avoiding Slow Ponsing.
;
; When static honsing is enabled, we activate an improvement to slow ponsing.
;
; In particular, if X and Y can be assigned addresses (see the discussion of
; Static Hons Addressing from hons-raw.lisp) without the use of an OTHER-HT or
; STR-HT, then we can just combine their addresses with hl-addr-combine (which
; is one-to-one) and use the resulting integer as our key. In many cases this
; allows us to avoid the hash table lookups required in Slow Ponsing.
;
; The basic ideas here are:
;
; - Some ACL2 objects (any static conses, symbol, or small fixnum) have
; addresses, but other objects (larger numbers, characters, strings, and
; ordinary conses) do not.
;
; - If (PONS X Y) is given X and Y that both have addresses, we basically
; just hash their addresses with hl-addr-combine. The resulting integer is
; used as the key.
;
; - Otherwise, we fall back to Slow Ponsing. Since slow ponsing always
; creates a cons instead of an integer, there's no possibility of confusion
; between the keys from the two schemes.
; Note that unlike much of the memoization code, pons address calculation can
; traffic in fixnums rather than mfixnums.
#+static-hons
(defun pons-addr-of-argument (x)
; Warning: *default-hs* needs to have been initialized before calling this
; function. As of this writing, pons-addr-of-argument is called only by
; pons-addr-hash, which is called only by pons, which calls
; (hl-maybe-initialize-default-hs) before calling pons-addr-hash.
; Warning: Keep this in sync with hl-addr-of and hl-addr-of-unusual-atom.
; See hl-addr-of. This is similar, except without the STR-HT or OTHER-HT we
; can simply fail to assign addresses to strings, large numbers, rationals, and
; so forth.
(cond ((eq x nil) 256)
((eq x t) 257)
((symbolp x)
(hl-symbol-addr x))
((and (typep x 'fixnum)
(<= hl-minimum-static-int (the fixnum x))
(<= (the fixnum x) hl-maximum-static-int))
(the fixnum
(+ hl-static-int-shift (the fixnum x))))
((characterp x)
(char-code x))
((and (consp x)
(hl-hspace-truly-static-honsp x *default-hs*))
; Note that a honsp is not garbage collected unless it is first freed from its
; addr-table. So the following value will correspond to x until a hons-wash or
; hons-clear is done, which is long enough since those operations clear all
; pons tables (in the #+static-hons case, which is the case we are in here).
; [Jared] Is this OK for multithreaded memoize? I think so. The scenario here
; is:
;
; - The memo tables are shared across all threads
; - Each thread has its own hons space and their own notion of which
; conses are hl-hspace-truly-static-honsp
; - It is NOT OK (as in, it would be a soundness bug) for us to keep memo
; table entries after hons spaces get washed or cleared, because the index
; we're computing could then be reassigned to some other cons, as described
; in the comment below about "mistake" and "contradiction".
;
; The current strategy is to have the hons wash/clear functions clear all the
; memo tables to ensure this doesn't happen. I think that's sufficient even in
; a multithreaded context: even though some honses in space A may not be honses
; in space B, they still have uniquely assigned indices (their hl-staticp
; indices), and that's all we're relying on here. In other words, I think we
; may get memoize misses because of differences in whether things are honsed,
; but we aren't going to get any false hits.
(+ hl-dynamic-base-addr (hl-staticp x)))
(t
; It is perhaps a bit tempting to add a case for (hl-staticp x), and then
; proceed as in the consp case of hl-addr-of. But if x is later garbage
; collected, then afterwards, each entry in the pons table created using this
; address is likely to be stale. In fact, such treatment of static conses
; allowed the following book to be certified in ACL2(h) Versions 6.5 and 5.0
; (and quite possibly all versions inbetween and maybe some earlier).
; (in-package "ACL2")
;
; (defn foo (x y) (+ (len x) (nfix y)))
;
; (memoize 'foo :recursive nil)
;
; (value-triple (progn$ (foo (hons-copy '(a)) 0)
; (foo (hons-copy '(b)) 0)
; (foo (hons-copy '(c)) 0)
; (foo (hons-copy '(d)) 0)
; (hons-wash)
; nil))
;
; (defthm mistake
; (equal (foo (hons-copy '(a b)) 0)
; 1)
; :hints (("Goal" :in-theory (e/d (foo (hons-copy))
; (hons-copy))))
; :rule-classes nil)
;
; (defthm contradiction
; nil
; :hints (("Goal" :use mistake))
; :rule-classes nil)
nil)))
; [Jared] multithreading notes. Pons tables are hash tables that will be
; shared. There are no locks in the pons code, because we assume that the lock
; will be grabbed before we start ponsing.
#+static-hons
(defabbrev pons-addr-hash (x y)
; Warning: *default-hs* needs to have been initialized before calling this
; function, because it calls pons-addr-of-argument. As of this writing,
; pons-addr-of-argument is called only by pons-addr-hash, which is called only
; by pons, which calls (hl-maybe-initialize-default-hs) before calling
; pons-addr-hash.
; We try to compute the addresses of X and Y and hash them together. If either
; doesn't have an address, we just return NIL.
(let ((xaddr (pons-addr-of-argument x)))
(if (not xaddr)
nil
(let ((yaddr (pons-addr-of-argument y)))
(if (not yaddr)
nil
(hl-addr-combine* xaddr yaddr))))))
(defmacro incf-pons-calls ()
'(safe-incf *pons-call-counter* 1))
(defmacro incf-pons-misses ()
'(safe-incf *pons-misses-counter* 1))
(defun pons (x y ht)
; We avoid any lock in this function, since it is called under the lock
; installed by with-global-memoize-lock-static in the memoized definition; see
; memoize-fn-def.
(declare (hash-table ht))
; With static honsing pons calls pons-addr-hash, which calls
; pons-addr-of-argument, which can access *default-hs*. So here, we make sure
; that *default-hs* is initialized with static honsing. This is inexpensive
; compared to pons, for two reasons:
; - When *default-hs* has already been initialized, the call just below is
; essentially just a read of a special variable.
; - Pons is only called on behalf of memoized functions with at least two
; arguments.
#+static-hons
(hl-maybe-initialize-default-hs)
#+static-hons
(let ((addr (pons-addr-hash x y)))
(when addr (return-from pons addr)))
(incf-pons-calls)
(let* ((flex-alist (mf-gethash y ht))
(entry (hl-flex-assoc x flex-alist)))
(cond
(entry)
(t
(incf-pons-misses)
(let* ((was-alistp (listp flex-alist))
(new-cons
; It might be tempting to think about creating a static cons here.
; But in order to benefit from that, we would need to find a good way
; to use its index, perhaps in place of new-cons as a key of the pons
; table. But in that case we would still need to hang on to that
; (static) cons so that it isn't garbage collected; so there is no
; obvious benefit to creating a static cons here, after all.
(cons x y))
(new-flex-alist (mf-flex-acons new-cons flex-alist)))
; Ctrl+C safety is subtle. If was-alistp, then the above was applicative. We
; now install the flex alist, which occurs as a single update to the hash
; table.
(when was-alistp
(mf-sethash y new-flex-alist ht))
; Otherwise, the flex-acons was non-applicative and the table was already
; extended, so there's nothing more we need to do.
new-cons)))))
(defmacro pist* (table &rest x)
(cond ((atom x) x)
((atom (cdr x)) (car x))
(t (list 'pons
(car x)
(cons 'pist* (cons table (cdr x)))
table))))
; SECTION: Identifying functions that are unsafe to memoize
(defun initialize-never-memoize-ht ()
; Our goal is to avoid memoizing functions that support the implementation of
; memoization. So we disallow functions defined in file hons-raw.lisp or in
; file memoize-raw.lisp, as well as built-ins Common Lisp functions. This
; might not be a complete list, and it might even be too comprehensive; but it
; seems like a reasonable start.
(let ((ht (mf-mht :test 'eq :shared :default)))
(note-fns-in-file "hons-raw.lisp" ht) ; its fns may be used in this file
(note-fns-in-file "memoize-raw.lisp" ht)
(flet ((update-nmht (fn)
(when (and (fboundp fn)
(not (macro-function fn)))
; We only memoize functions, so we can restrict according to the test above.
(never-memoize-fn fn))))
(mf-maphash (lambda (key val)
(declare (ignore val))
(update-nmht key))
ht)
(loop for fn in *common-lisp-symbols-from-main-lisp-package*
do
(update-nmht fn))))
nil)
(defun never-memoize-fn (fn)
; Warning: Keep the return values in sync for the logic and raw Lisp.
; Note: Consider locking uses of this function, since it updates a global
; hash-table.
(mf-sethash fn t *never-memoize-ht*)
nil)
(defun never-memoize-p (fn)
(mf-gethash fn *never-memoize-ht*))
; SECTION: Preliminary memoization support
(defrec memoize-info-ht-entry
; The *memoize-info-ht* associates memoized function symbols with these
; records.
; The fields are vaguely ordered by most frequently referenced first, but it's
; not clear that this is important for efficiency, except perhaps for
; ext-anc-attachments (see update-memo-entry-for-attachments).
(ext-anc-attachments ; see the Essay on Memoization with Attachments
; Start-ticks is a symbol whose value is the start time (in float ticks) of the
; current, outermost call of fn, or -1 if no call of fn is in progress.
start-ticks ; see above
num ; an integer, unique to fn
tablename ; a symbol whose value is the memoize table for fn
ponstablename ; a symbol whose value is the pons table for fn
condition ; :condition passed to memoize-fn (but nil for *nil*)
inline ; T or NIL :inline arg, as passed to memoize-fn
memoized-fn ; the new value of (symbol-function fn)
old-fn ; the old value of (symbol-function fn), or nil
fn ; a symbol, the name of the function being memoized
sts ; the stobj memo-table lists for fn
; Cl-defun is the function body actually used, in the inline=t case, as
; supplied (or as computed, if not supplied).
cl-defun ; see above
formals ; as supplied (or as computed, if not supplied)
commutative ; asserts this is a binary commutative function
specials ; hack for raw Lisp functions that read specials
stobjs-in ; as supplied (or as computed, if not supplied)
stobjs-out ; as supplied (or as computed, if not supplied)
record-bytes ; value as bound at the time MEMOIZE-FN is called
record-calls ; ''
record-hits ; ''
record-mht-calls ; ''
record-pons-calls ; ''
record-time ; ''
forget ; Boolean, clears memo when outermost call exits
memo-table-init-size ; integer, reasonable default of *mht-default-size*
aokp ; use of attachments is allowed
; [Jared] multithreading notes: we could probably add a lock here that was only
; on a per-function basis. That could help avoid needing the global lock, much
; of the time.
)
t)
(defmacro ma-index (col &optional (row 0) (2mmf '*2max-memoize-fns*))
; This function creates an index into the one-dimensional array,
; *memoize-call-array*. That array is conceptually a two-dimensional array,
; and the arguments are a corresponding column and row of that 2D array. If
; row is not supplied, then the index returned points to the start of the
; column (i.e., for a row of 0).
(let ((col-base `(* ,2mmf
(the-mfixnum ,col))))
(if (eql row 0)
`(the-mfixnum ,col-base)
`(ma-index-from-col-base ,col-base ,row))))
(defmacro ma-index-from-col-base (col-base row)
; This function creates an index into the one-dimensional array,
; *memoize-call-array*. That array is conceptually a two-dimensional array,
; and the arguments are a corresponding column and row of that 2D array.
; Col-base is an index into *memoize-call-array* representing the start of a
; column, and row represents a row of the 2D array.
`(the-mfixnum (+ (the-mfixnum ,col-base)
(the-mfixnum ,row))))
(defmacro ma-index-calls (child &optional
(parent 0)
(2mmf '*2max-memoize-fns*))
; When parent calls child, we will make an entry in *memoize-call-array* for
; the "column" of parent at a "row" corresponding to child, where "column" and
; "row" are in quotes because while we think of *memoize-call-array* as a
; two-dimensional array, it actually has only one dimension. This macro
; computes the index into *memoize-call-array* for that entry. A special case
; is "column" 0 of *memoize-call-array*, which is used as scratch space by
; COMPUTE-CALLS-AND-TIMES for sums across all functions. In that case there is
; no "parent" -- we are just accumulating all calls regardless of any notion of
; "parent".
(let ((2child `(the-mfixnum (* 2 (the-mfixnum ,child)))))
(if (eql parent 0)
2child ; simplification of `(ma-index 0 ,2child ,2mmf)
`(ma-index ,parent ,2child ,2mmf))))
(defmacro ma-index-ticks (child &optional
(parent 0)
(2mmf '*2max-memoize-fns*))
; This macro is similar to ma-index-calls, but it counts ticks instead of
; calls. See comments in ma-index-calls.
(let ((2child+1 `(the-mfixnum
(1+ (the-mfixnum (* 2 (the-mfixnum ,child)))))))
(if (eql parent 0)
2child+1 ; simplification of `(ma-index 0 ,2child+1 ,2mmf)
`(ma-index ,parent ,2child+1 ,2mmf))))
(defun outside-caller-col-base ()
; See *caller*. Notice that although we pass *initial-max-symbol-to-fixnum*
; explicitly below, the value also depends on the global *2max-memoize-fns*.
(ma-index *initial-max-symbol-to-fixnum*))
(defvar *caller*
; This variable represents the index into *memoize-call-array* that starts the
; layout of the "column" for the innermost memoized function call being
; evaluated (or, in the case of its initial value, the column for
; "outside-caller" = (mf-gethash *initial-max-symbol-to-fixnum*
; *memoize-info-ht*). The idea is to accumulate calls and ticks into that
; "column".
; When memoized functions are executing in parallel, the value of *caller* and
; of statistics derived therefrom may be meaningless and random.
; At one time we introduced this variable using defg instead of defvar. To see
; why we now use defvar, see the comment about unwind-protect in
; memoize-fn-outer-body.
(let ((val (outside-caller-col-base)))
(check-type val mfixnum)
val))
(declaim (type mfixnum *caller*))
(defrec memo-max-sizes-entry
; This represents a single entry in the *memo-table-max-sizes* table.
(num-clears ; how many times has this table been cleared (nat)
max-pt-size ; maximum size of the pons table before any clear (nat)
max-mt-size ; maximum size of the memo table before any clear (nat)
avg-pt-size ; average size of the pons table before any clear (float)
avg-mt-size ; average size of the memo table before any clear (float)
)
t)
(defun make-initial-memoize-hash-table (fn init-size)
; Fn is the name of a function. Init-size is the initial size that the user
; says we should use. We want to create and return a new hash table for this
; function's memoization table. One possible implementation of this function
; would just be:
;
; (mf-mht :size init-size)
;
; But we hope to do better. Our idea is to look at how large the table has
; been in the past, and use that size to make a good prediction of how large
; the table will be this time.
;
; The idea here is to build a table that's just slightly bigger than the
; average size we've seen so far. We arbitrarily say that "slightly bigger"
; means 1.2x the previous average.
;
; By itself this would be scary. Big hash tables can use a lot of memory. In
; early versions of this code we thought that 1 MB of space provided about
; 44,000 entries. But here we see in CCL that 1 MB of space can buy you about
; 53,000 entries:
; ? [RAW LISP] (time (make-hash-table :size 53102))
; (MAKE-HASH-TABLE :SIZE 53102)
; took 606 microseconds (0.000606 seconds) to run.
; During that period, and with 4 available CPU cores,
; 453 microseconds (0.000453 seconds) were spent in user mode
; 126 microseconds (0.000126 seconds) were spent in system mode
; 999,984 bytes of memory allocated.
; #<HASH-TABLE :TEST EQL size 0/53102 #x30200276BDDD>
; ? [RAW LISP]
; We want to avoid creating a hundred-megabyte memo table for a function just
; because it was used heavily for a short while and then cleared once before.
; On the other hand, if a memo table truly does get large on a regular basis,
; then we do want to guess a big size for it.
;
; So in this code, we enforce an artificial maximum on our guess, but allow
; this maximum to grow with the number of times we've cleared the table.
; Basically we're comfortable allowing the maximum guess to grow at a rate of 1
; MB per clear. If a table has been cleared 100 times, we have a pretty good
; sense of its average usage and we can be comfortable allocating up to 100 MB
; for it. If it's been cleared more than 1000 times, the cap is a gigabyte.
; But of course, to actually reach such a large guess, you'd have to be
; repeatedly filling up the table to contain millions of entries and then
; clearing it.
; Note: The 44,000 in the code below stems from our earlier estimate that it's
; equivalent to 1 MB. We have no data to suggest this slightly smaller growth
; rate (.83 MB instead of 1 MB) hurts us and so we have left it.
(with-global-memoize-lock
(let* ((max-sizes ; previously recorded sizes of this table, if any exist
(mf-gethash fn *memo-max-sizes*))
(size-to-use
(if (not max-sizes)
; We never cleared this memoize table before, so we don't have anything to go
; on besides what the user says. Do what they say.
init-size
(let* ((nclears (access memo-max-sizes-entry max-sizes
:num-clears))
(avg-mt-size (access memo-max-sizes-entry max-sizes
:avg-mt-size))
(our-guess (ceiling (* 1.20 avg-mt-size)))
(capped-guess (min our-guess (* nclears 44000)))
(final-guess (max 60 init-size capped-guess)))
final-guess))))
(mf-mht
; This table is essentially blown away and reconstituted by rememoize-all,
; hence by (mf-multiprocessing t). So we avoid using :shared :default below,
; in order to make the table a bit more efficient in the uniprocessing case.
:size size-to-use))))
(defun make-initial-memoize-pons-table (fn init-size)
(declare (ignorable init-size))
; This is similar to make-initial-memoize-table, but for the pons table.
(with-global-memoize-lock
(let* ((max-sizes (mf-gethash fn *memo-max-sizes*))
(size-to-use
(if (not max-sizes)
; We've never cleared this pons table before, so we don't have anything to go
; on besides what the user says. Now, this is subtle. Originally we just
; returned init-size here, i.e., "do what the user says." But while this makes
; sense for the memo table, it doesn't necessarily make much sense for the pons
; table. In particular, we can sometimes avoid ponsing by using our
; static-cons-index-hashing scheme.
; In some sense it would probably be good to give the user explicit control
; over the pons table size. But for now, the main use of our memoize table
; size controls is to set things up for big BDD/AIG/SEXPR operations where
; we've got honsed data. So, we just use *mht-default-size* here, which seems
; as reasonable a choice as any, so that the memo-table-init-size argument of
; memoize-fn only affects the memoize table and not the initial pons table.
*mht-default-size*
(let* ((nclears (access memo-max-sizes-entry max-sizes
:num-clears))
(avg-pt-size (access memo-max-sizes-entry max-sizes
:avg-pt-size))
(our-guess (ceiling (* 1.20 avg-pt-size)))
(capped-guess (min our-guess (* nclears 44000)))
(final-guess (max *mht-default-size*
init-size
capped-guess)))
final-guess))))
(mf-mht
; This table is essentially blown away and reconstituted by rememoize-all,
; hence by (mf-multiprocessing t). So we avoid using :shared :default below,
; in order to make the table a bit more efficient in the uniprocessing case.
:size size-to-use))))
(defun update-memo-max-sizes (fn pt-size mt-size)
; Fn is a memoized function with a non-nil pons-table or memo-table, which is
; about to receive a new pons-table and memo-table of size pt-size and mt-size
; respectively (perhaps nil initially, in which case the size is 1). We update
; the *memo-max-sizes* entry associated with f. The :max-pt-size and
; :max-mt-size are not used any more; they were used by Jared to track how big
; the tables were growing.
(with-global-memoize-lock
(let ((old (mf-gethash fn *memo-max-sizes*)))
(if (not old)
(mf-sethash fn
(make memo-max-sizes-entry
:num-clears 1
:max-pt-size pt-size
:max-mt-size mt-size
:avg-pt-size (coerce pt-size 'float)
:avg-mt-size (coerce mt-size 'float))
*memo-max-sizes*)
(let* ((old.num-clears (access memo-max-sizes-entry old :num-clears))
(old.max-pt-size (access memo-max-sizes-entry old :max-pt-size))
(old.max-mt-size (access memo-max-sizes-entry old :max-mt-size))
(old.avg-pt-size (access memo-max-sizes-entry old :avg-pt-size))
(old.avg-mt-size (access memo-max-sizes-entry old :avg-mt-size))
(new.num-clears (+ 1 old.num-clears)))
(mf-sethash fn
(make memo-max-sizes-entry
:num-clears new.num-clears
:max-pt-size (max pt-size old.max-pt-size)
:max-mt-size (max mt-size old.max-mt-size)
:avg-pt-size (/ (+ pt-size (* old.avg-pt-size
old.num-clears))
new.num-clears)
:avg-mt-size (/ (+ mt-size (* old.avg-mt-size
old.num-clears))
new.num-clears))
*memo-max-sizes*)))))
nil)
(defun print-memo-max-sizes ()
; Note (Matt K., 9/2014): This is just a printing function, probably rarely
; used, so I haven't really reviewed this code, though I did test it
; successfully.
(with-global-memoize-lock
(when (equal (mf-hash-table-count *memo-max-sizes*) 0)
(return-from print-memo-max-sizes nil))
(format t "Memo table statistics gathered at each from when they were ~
cleared:~%~%")
(let ((indent 8) ; initial value of 8 is the length of "Function"
indent-str)
(mf-maphash (lambda (fn entry)
(declare (ignore entry))
(setq indent (max indent (length (symbol-name fn)))))
*memo-max-sizes*)
(setq indent-str (format nil "~a" (+ 2 indent)))
(format t (concatenate 'string "~" indent-str ":@a") "Function")
(format t " ~10:@a | ~15:@a ~15:@a | ~15:@a ~15:@a~%"
"Clears" "PT Max" "PT Avg" "MT Max" "MT Avg")
(mf-maphash
(lambda (fn entry)
(let* ((num-clears (access memo-max-sizes-entry entry :num-clears))
(max-pt-size (access memo-max-sizes-entry entry :max-pt-size))
(max-mt-size (access memo-max-sizes-entry entry :max-mt-size))
(avg-pt-size (access memo-max-sizes-entry entry :avg-pt-size))
(avg-mt-size (access memo-max-sizes-entry entry :avg-mt-size)))
(format t
(concatenate 'string "~" indent-str
":@a ~10:D | ~15:D ~15:D | ~15:D ~15:D~%")
fn num-clears
max-pt-size (floor avg-pt-size)
max-mt-size (floor avg-mt-size))))
*memo-max-sizes*)
(format t "~%")))
nil)
; Essay on Memoization Involving Stobjs
; We allow memoization of functions that take user-defined stobjs (not state)
; as arguments but do not return stobjs. The key is the use of memoize-flush
; to "forget" all that was remembered for certain functions that use certain
; stobjs. We must keep memoize-flush very fast in execution so as not to slow
; down stobj update or resize operations in general. Indeed, memoize-flush may
; (according to tests run) incur essentially no cost (after Version_4.3) as
; long as no functions with stobj arguments are actually memoized.
; The following example shows why we disallow memoization of functions that
; return stobjs. First, redefine memoize-table-chk by eliminating the branch
; that causes an error in the presence of stobj names in stobjs-out. Then
; start up ACL2 and submit the forms below. The problem is that we do not
; inhibit storing a result in the case that the stobj has changed from the time
; the function was called to the time the result is to be stored.
; (defstobj st fld)
; (defun foo (st)
; (declare (xargs :stobjs st))
; (let ((st (update-fld (cons (fld st) (fld st)) st)))
; (mv (fld st) st)))
; (foo st) ; updates (fld st), returns (mv (nil) st)
; (memoize 'foo)
; (foo st) ; updates (fld st), returns (mv ((nil) nil) st)
; (foo st) ; no longer updates (fld st)
; (foo st) ; no longer updates (fld st)
; (fld st) ; still ((nil) nil), whereas if we hadn't
; ; memoized foo (fld st) would be:
; ; ((((NIL) NIL) (NIL) NIL) ((NIL) NIL) (NIL) NIL)
(defun memoize-flush1 (lst)
; Lst is the list of names of memoize tables, typically the tables associated
; with all the memoized stobj-using (but not stobj-returning) functions on a
; given stobj st.
; Experiments showed that when lst is nil, it is faster to call this function
; than to inline its code into the body of memoize-flush.
; We "forget" all memoized values by clearing all necessary memoize tables; see
; the comment about memoize-flush in memoize-fn. We leave the pons table alone
; in order to keep this flushing operation as fast as possible. Note that the
; pons table merely stores keys to be looked up in the memo table, so there is
; no soundness issue, and in fact those pons table entries might remain useful;
; the cost is the space taken up by the pons tables.
(with-global-memoize-lock
(loop for sym in lst do
(when (boundp (the symbol sym)) ; not sure if this test is needed
(let ((old (symbol-value (the symbol sym))))
(unless (or (null old)
(eql 0 (mf-hash-table-count old)))
(setf (symbol-value (the symbol sym)) nil)))))))
(defmacro memoize-flush (st)
; St is a stobj name and (st-lst st), below, returns returns a symbol whose
; value is a list in which are saved the names of the memoize tables that will
; be set to nil whenever the stobj st is changed.
; See memoize-flush1 for a relevant discussion.
(and st
(let ((s (st-lst st)))
`(when ,s ; optimization
(memoize-flush1 ,s)))))
(defun sync-memoize-call-array ()
; Warning: Be careful if you call this function from other than memoize-init or
; memoize-call-array-grow. In particular, it is probably a mistake to call
; this function while memoized functions are running, since *caller* is
; supposed to depend on the innermost memoized function call being evaluated,
; but sync-memoize-call-array sets *caller* to denote the "outside" caller.
; We assume that *2max-memoize-fns* is the intended length of
; *memoize-call-array* and *max-symbol-to-fixnum* is the maximum fixnum
; associated with a memoized function symbol (so far).
; It should be rare that this function is called after its call by
; memoize-init, i.e., by a call from memoize-call-array-grow. That is:
; memoize-call-array-grow should be rare. This code historically often cleared
; *memoize-call-array* and *callers-array* in most cases when called by
; memoize-call-array-grow. For simplicity, we make that be the case always,
; and clearly.
(with-global-memoize-lock
(let ((n1 (ma-index *2max-memoize-fns*))
(n2 (1+ *max-symbol-to-fixnum*)))
(declare (type (and fixnum mfixnum) n1 n2))
(setq *memoize-call-array* (initial-memoize-call-array 1))
(setq *callers-array* (make-array 0))
(gc$)
(setq *memoize-call-array* (initial-memoize-call-array n1))
(setq *callers-array* (make-array n2 :initial-element nil))
(setq *caller* (outside-caller-col-base)))))
(defun memoize-call-array-grow
(&optional (2nmax (* 2 (ceiling (* 3/2 (/ *2max-memoize-fns* 2))))))
; In our own code we call this function with no arguments. However, as of this
; writing, community book file centaur/memoize/old/profile-raw.lsp calls it
; with an argument. The argument will become the new dimension of the
; (virtual) 2D square array, *memoize-call-array*.
(with-global-memoize-lock
(unless (integerp 2nmax)
(error "(memoize-call-array-grow ~s). Arg must be an integer."
2nmax))
(unless (evenp 2nmax)
(error "(memoize-call-array-grow ~s). Arg must be even." 2nmax))
(unless (> 2nmax 100)
(error "(memoize-call-array-grow ~s). Arg must be > 100." 2nmax))
(when (<= 2nmax *2max-memoize-fns*)
(cw "; memoize-call-array-grow: *memoize-call-array* already big enough.~%")
(return-from memoize-call-array-grow))
(unless (<= (* 2nmax 2nmax) most-positive-mfixnum)
(error "memoize-call-array-grow: most-positive-mfixnum exceeded. Too ~
many memoized functions."))
(unless (<= (* 2nmax 2nmax) most-positive-fixnum)
(error "memoize-call-array-grow: most-positive-fixnum exceeded. Too ~
many memoized functions."))
(unless (< (* 2nmax 2nmax) array-total-size-limit)
(error "memoize-call-array-grow: ARRAY-TOTAL-SIZE-LIMIT exceeded. Too ~
many memoized functions."))
(unless (< (* 2nmax 2nmax) array-dimension-limit)
(error "memoize-call-array-grow: ARRAY-DIMENSION-LIMIT exceeded. Too ~
many memoized functions."))
(unless (eql *caller* (outside-caller-col-base))
; Ouch. We are about to blow away and reconsitute the memoization structures,
; including memoized definitions, yet we are inside a memoized definition!
; This seems like a potentially terrible situation; presumably the original
; definition continues executing in this case even though the function has been
; redefined, and probably the only thing wrong is the statistics gathering.
; But are we sure? Fortunately, we are only here when we need more than the
; initial limit of about 500 memoized functions, and it seems unlikely that
; this will happen when inside a memoized function. (It might happen when
; calling profile-all or profile-acl2 in raw Lisp, as defined in community book
; books/centaur/memoize/old/profile.lisp. But it seems unlikely that these
; would be called inside some other memoized function!)
; So we cause a somewhat scary continuable error here, and leave it up to the
; user whether to continue at his or her own risk. If this problem really
; occurs in practice, someone can point it out, and some attention can perhaps
; be given to the problem of whether there is any reasonable way to guarantee
; that all is well in this circumstance.
(cerror "Continue at your own risk."
"MEMOIZE-CALL-ARRAY-GROW was called while a memoized function~%~
was executing, so all bets are off. Continue only at your own~%~
risk. That risk may be low, but at the least, memoization~%~
reports (in particular, from memsum) will probably be~%~
inaccurate."))
(setq *2max-memoize-fns* 2nmax)
(let ((state *the-live-state*))
(observation 'memoize-call-array-grow
"Now reinitializing memoization structures. This will ~
erase saved values and statistics."))
(sync-memoize-call-array)
(rememoize-all))
nil)
(defun-one-output symbol-to-fixnum-create (s)
; We return the number n associated with function symbol s, in the sense that
; (mf-gethash n *memoize-info-ht*) = s. If no such number n yet exists, the
; next available number is returned -- but this function does not actually add
; the association; that is done by its caller, memoize-fn.
(check-type s symbol)
(with-global-memoize-lock
(let ((g (mf-gethash s *memoize-info-ht*)))
(if g
(access memoize-info-ht-entry g :num)
(let (new)
(loop for i fixnum
from (1+ *initial-max-symbol-to-fixnum*)
below (the fixnum (floor *2max-memoize-fns* 2))
do (unless (mf-gethash i *memoize-info-ht*)
(setq new i)
(return)))
(cond (new (setq *max-symbol-to-fixnum*
(max *max-symbol-to-fixnum* new))
new)
(t (memoize-call-array-grow)
; Everything has been blown away! So try again.
(symbol-to-fixnum-create s))))))))
(defun-one-output symbol-to-fixnum (s)
; This is like symbol-to-fixnum-create, but assumes that s is already memoized.
(check-type s symbol)
(let ((g (mf-gethash s *memoize-info-ht*)))
(if g
(access memoize-info-ht-entry g :num)
(error "(symbol-to-fixnum ~s). Illegal symbol." s))))
(defun-one-output fixnum-to-symbol (n)
(check-type n fixnum)
(or (mf-gethash n *memoize-info-ht*)
(error "(fixnum-to-symbol ~s). Illegal number." n)))
(defun-one-output hons-gentemp (root)
(check-type root string)
(loop
(safe-incf *hons-gentemp-counter* 1 hons-gentemp)
(let ((name (our-syntax
(format nil "HONS-G-~s,~s" root *hons-gentemp-counter*))))
(multiple-value-bind (sym status)
(intern name (find-package "ACL2_INVISIBLE"))
(if (null status) (return sym))))))
(defun-one-output st-lst (st)
; ST-LST returns a symbol whose value is a list in which are saved the
; names of the memoize tables that will be set to nil whenever the
; stobj st is changed.
(check-type st symbol)
(intern (our-syntax (format nil "HONS-S-~s,~s"
(package-name (symbol-package st))
(symbol-name st)))
(find-package "ACL2_INVISIBLE")))
(defun-one-output mf-dcls (l)
; Drop any (IGNORE ...) from within a list of declarations, ((DECLARE ...)
; (DECLARE ...) ...) and return the resulting list of DECLAREs.
(loop for dec in l nconc
(let ((temp
(if (consp dec)
(loop for d in (cdr dec) nconc
(if (and (consp d) (eq (car d) 'ignore))
nil ; ignorable is added in memoize-fn-def
(cons d nil))))))
(if temp (cons (cons 'declare temp) nil)))))
(defun memoize-use-attachment-warning (fn at-fn)
(when *memoize-use-attachment-warning-p*
(let ((state *the-live-state*))
(warning$ 'top-level "Attachment"
"Although the function ~x0 is memoized, a result is not being ~
stored because ~@1. Warnings such as this one, about not ~
storing results, will remain off for all functions for the ~
remainder of the session unless the variable ~x2 is set to a ~
non-nil value in raw Lisp."
fn
(mv-let (lookup-p at-fn)
(if (consp at-fn)
; Then at-fn is of the form (:lookup . f), where f was memoized with :aokp t
; and a saved value for f was used, as explained in the lookup-p case below.
(assert$ (eq (car at-fn) :lookup)
(mv t (cdr at-fn)))
(mv nil at-fn))
(cond (lookup-p
(msg "a stored result was used from a call of ~
memoized function ~x0, which may have ~
been computed using attachments"
at-fn))
(t
(msg "an attachment to function ~x0 was used ~
during evaluation of one of its calls"
at-fn))))
'*memoize-use-attachment-warning-p*))
(setq *memoize-use-attachment-warning-p* nil)))
(defun-one-output memoize-fn-suffix (str sym)
(check-type str string)
(check-type sym symbol)
(let ((spkn (package-name (symbol-package sym)))
(sn (symbol-name sym)))
; Sometimes we use our-syntax for calls like this, and sometimes we don't.
; It's not clear that it much matters, so we have tended to preserve what was
; true historically.
(format nil "~s,~s,~s" str spkn sn)))
(defmacro mf-index (x &optional known-not-rational)
(assert (symbolp x)) ; else we should use defabbrev here
`(cond ,@(unless known-not-rational
`(((typep ,x 'fixnum) ,x)))
#+static-hons
(t
; Since this function is only heuristic, it is sound to use the weak staticp
; check below, instead of using consp and hl-hspace-truly-static-honsp checks
; as we do in pons-addr-of-argument. We are hopeful that this decision won't
; result in expensive needless misses when checking memoization tables, but
; perhaps more thought about that would be useful here.
(hl-staticp ,x))))
(defun-one-output mis-ordered-commutative-args (x y)
; This "function" need not return the same value for every call on the same
; input argument list. We use this function in the creation of a suitable
; memoization table key, specifically to decide whether to reorder x and y
; before calling pist* (see memoize-fn). Such reordering is justified for an
; ACL2 function of two arguments that has been proven commutative, because such
; reordering causes no difference in the result that can be detected by the
; ACL2 logic.
; The basic idea is to create an "index" for each argument. For static conses
; we use hl-staticp as the index; but some Lisps have no static conses. Of
; course, it is always legal to return nil. Perhaps in the future we will use
; machine addresses or some other technique for sorting the arguments.
; We order objects that are rational or have an index before objects that are
; neither.
(cond
((eql x y) nil)
(t (let ((idx (mf-index x)))
(declare (type (or null fixnum) idx))
(cond
(idx
(let ((idy (mf-index y)))
(declare (type (or null fixnum) idy))
(cond
(idy (< (the fixnum idy)
(the fixnum idx)))
((rationalp y)
(< y (the fixnum idx))))))
((rationalp x)
(let ((idy (mf-index y)))
(declare (type (or null fixnum) idy))
(cond
(idy (< (the fixnum idy)
x))
((rationalp y)
(< y x)))))
((or (rationalp y)
(mf-index y t))))))))
(defun memoize-look-up-def (fn cl-defun inline wrld
&key quiet-p)
; This function returns a definition (fn formals ...), that is, without the
; initial DEFUN. It returns cl-defun unchanged (other than stripping off the
; car if it is DEFUN), unless cl-defun is :DEFAULT -- which won't happen when
; memoize is called in the ACL2 loop (search in the definition of add-trip for
; the form (assert$ cl-defun (memoize-fn (nth 1 tuple) ...)). In that :DEFAULT
; case, a reasonable attempt is made to find a definition.
; In memoize-fn, we rely on the fact that when memoize-look-up-def is called
; with a non-nil value of inline and it returns without error, then it returns
; a non-nil definition (which is a true list).
(cond ((eq cl-defun :default)
(cond
((null inline) nil)
((not (fboundp fn))
(error "MEMOIZE-LOOK-UP-DEF: ** ~a is undefined." fn))
((let ((def (cltl-def-from-name fn wrld)))
(cond (def (assert (eq (car def) 'defun))
(cdr def)))))
(t (memoize-look-up-def-raw fn quiet-p))))
((eq (car cl-defun) 'defun)
(cdr cl-defun))
(t
cl-defun)))
(defg *memoize-timing-bugs* 0)
(defmacro fix-ticks (ticks ctx)
; Warning: If you make this a function and add a declaim form first such as
; (declaim (ftype (function (mfixnum t) (values mfixnum))
; fix-ticks))
; then in GCL, make sure that after the build, the symbol-plist shows the
; correct declaim information. We have seen a version of GCL 2.6.12pre for
; which this was not the case, which apparently was the cause of an unexpected
; error.
`(let ((ticks ,ticks)
(ctx ,ctx))
(declare (type mfixnum ticks))
(the-mfixnum
(cond
((and (<= 0 (the-mfixnum ticks))
(< (the-mfixnum ticks)
(the-mfixnum *10-days*)))
(the-mfixnum ticks))
(t
(incf *memoize-timing-bugs*)
(when (<= *memoize-timing-bugs* 10)
(format t "Ignoring time increment of ~a sec for ~a~%"
(/ ticks *float-ticks/second*) ctx)
(when (eql *memoize-timing-bugs* 10)
(format
t
"Timing is very dubious. Suppressing further messages.~%")))
0)))))
; SECTION: The main code for creating memoized function definitions:
; memoize-fn, unmemoize-fn, etc.
(defun-one-output memoizedp-raw (fn)
(and (symbolp fn)
(values (mf-gethash fn *memoize-info-ht*))))
(defmacro mf-make-symbol (&rest r) ; for forming symbols in package ACL2
`(our-syntax (intern (format nil ,@r) *acl2-package*)))
(defun memoize-fn-init (fn inline wrld &aux (state *the-live-state*))
; Do initial stuff for memoize-fn, including error checking. Quite possibly
; only a few of the checks below (if any) are necessary if memoize-fn is called
; by way of an invocation of memoize in the ACL2 loop.
(with-global-memoize-lock
(unless (symbolp fn)
(error "Memoize-fn: ~s is not a symbol." fn))
(maybe-untrace! fn) ; See the comment about Memoization in trace$-def.
(unless *memoize-init-done*
(error "Memoize-fn: *MEMOIZE-INIT-DONE* is still nil."))
(unless (fboundp fn)
(error "Memoize-fn: ~s is not fboundp." fn))
(when (or (macro-function fn)
(special-operator-p fn)
(and (fboundp 'compiler-macro-function) ; for GCL as of 5/2013
(compiler-macro-function fn)))
(error "Memoize-fn: ~s is a macro or a special operator or has a ~
compiler macro."
fn))
(when (never-memoize-p fn)
(error "Memoize-fn: ~s must never be memoized" fn))
(when (and inline
(or (member-eq fn
(f-get-global 'logic-fns-with-raw-code state))
(member-eq fn
(f-get-global 'program-fns-with-raw-code state))))
(error "Memoize-fn: ~s must never be memoized with :inline t,~%~
because it belongs to the list stored in state global~%~
variable ~s and thus runs code different~%~
than that stored in the ACL2 logical world, as is typically~%~
used when inserting inlined code for memoization."
fn
(if (member-eq fn
(f-get-global 'logic-fns-with-raw-code state))
'logic-fns-with-raw-code
'program-fns-with-raw-code)))
(when (memoizedp-raw fn)
(observation 'memoize
"~s0 is currently memoized. We will first unmemoize it, ~
then memoize it again."
fn)
(unmemoize-fn fn))
(when (member fn (eval '(trace)))
(observation 'memoize
"Untracing ~s before memoizing it."
fn)
(eval `(untrace ,fn)))
; TRACE, UNTRACE, OLD-TRACE, and OLD-UNTRACE are macros that get
; redefined sometimes. So we use EVAL in calling them.
#+ccl
(when (ccl::%advised-p fn)
(error "~%; Memoize-fn: Please unadvise ~s before calling memoize-fn on ~
it." fn))
(when (and (fboundp 'old-trace)
(member fn (eval '(old-trace))))
(observation 'memoize
"Old-untracing ~s before memoizing it."
fn)
(eval `(old-untrace ,fn)))
(when (getprop fn 'constrainedp nil 'current-acl2-world wrld)
(error "Memoize-fn: ~s is constrained; you may instead wish to memoize a ~
caller or to memoize its attachment (see :DOC defattach)."
fn))
#+ccl
(when (multiple-value-bind (req opt restp keys)
(ccl::function-args (symbol-function fn))
(or restp
keys
(not (integerp req))
(not (eql opt 0))))
; This sanity check will never cause an error when memoization is invoked using
; memoize inside the ACL2 loop.
; CCL documentation returned by
; (documentation 'ccl::function-args 'function)
; justifies the four bound variables above. The additional 5 values returned
; by ccl::function-args are discarded, as per the CL HyperSpec documentation of
; multiple-value-bind.
(error "Memoize-fn: ~a has non-simple arguments." fn))))
(defun memoize-fn-init-2 (fn formals condition stobjs-in stobjs-out)
; Do more initial stuff for memoize-fn after formals is computed.
(unless (and (symbol-listp formals)
(no-duplicatesp formals)
(loop for x in formals never (constantp x)))
; This sanity check is unnecessary when memoization is invoked using memoize
; inside the ACL2 loop.
(error "Memoize-fn: FORMALS, ~a, must be a true list of distinct, ~
nonconstant symbols."
formals))
(when (intersection lambda-list-keywords formals)
; This sanity check is unnecessary when memoization is invoked using memoize
; inside the ACL2 loop.
(error "Memoize-fn: FORMALS, ~a, may not intersect LAMBDA-LIST-KEYWORDS."
formals))
(when (and condition (or (member 'state stobjs-in)
(member 'state stobjs-out)))
; This sanity check is unnecessary when memoization is invoked using memoize
; inside the ACL2 loop, because in memoize-table-chk, we disallow state as a
; formal and we disallow every stobj as an output.
; Although we don't support memoization of functions involving state, we do
; allow user-defined stobjs as inputs, by laying down code in stobj-let-fn-raw
; and defstobj-field-fns-raw-defs to flush memo tables as necessary. But there
; is no such accommodation for the ACL2 state.
(error "Memoize-fn: ~s uses STATE." fn)))
(defun memoize-fn-inner-body (fn condition body fn-col-base
tablename localtablename
ponstablename localponstablename
memo-table-init-size
formals specials stobjs-out number-of-args
commutative aokp prog1-fn)
; This is the main part of the body of the function defined by memoize-fn.
; Comments saying "performance counting*" are intended to mark code that might
; best be ignored on a first reading. A long comment in *memoize-call-array*
; outlines how performance counting is implemented.
;; [Jared] multithreading note: all of this is going to be protected with the
;; global lock, so we shouldn't need any locking inside here. See
;; memoize-fn-def.
(let ((mf-record-mht ; performance counting
(and *record-mht-calls*
`((safe-incf
(aref ,*mf-ma*
,(ma-index-from-col-base fn-col-base *ma-mht-index*))
1))))
(mf-record-hit ; performance counting
(and *record-hits*
condition ; never record hits when profiling
`((safe-incf
(aref ,*mf-ma*
,(ma-index-from-col-base fn-col-base *ma-hits-index*))
1)))))
`(cond
,@(and (not (member-equal condition '('t t))) ; minor optimization
`(((not ,condition)
; See the comment in the definition of *condition-nil-as-hit*.
,@(and *condition-nil-as-hit*
mf-record-hit)
,body)))
,@(and
condition ; else dead code
`((t
(let (,*mf-ans* ,*mf-args* ,*mf-ans-p*)
(declare (ignorable ,*mf-ans* ,*mf-args* ,*mf-ans-p*))
(when (null ,tablename) ; e.g., after (clear-memoize-tables)
,@mf-record-mht
(setq ,tablename (make-initial-memoize-hash-table
',fn ,memo-table-init-size))
,@(and (> number-of-args 1)
`((setq ,ponstablename
(make-initial-memoize-pons-table
',fn ,memo-table-init-size)))))
(setq ,localtablename ,tablename)
,@(and (> number-of-args 1)
`((assert ,ponstablename)
(setq ,localponstablename ,ponstablename)))
; Generate the pons key. If there is just one arg, pist* just returns the arg
; and doesn't do any ponsing.
(setq ,*mf-args*
,(cond
(commutative
`(cond ((mis-ordered-commutative-args
,(car formals)
,(cadr formals))
(pist* ,localponstablename
,(cadr formals)
,(car formals)
,@specials))
(t
(pist* ,localponstablename
,(car formals)
,(cadr formals)
,@specials))))
(t `(pist* ,localponstablename
,@formals
,@specials))))
; Attempt to find the saved result.
(multiple-value-setq
(,*mf-ans* ,*mf-ans-p*)
,(let ((gethash-form `(mf-gethash ,*mf-args* ,localtablename)))
(cond (aokp `(cond
(*aokp* ,gethash-form)
(t (values nil nil))))
(t gethash-form))))
(cond
(,*mf-ans-p*
; Memoized lookup succeeded. Update performance counts and return result.
,@(and aokp
; We note that a saved value for fn was used, which may have depended on an
; attachment since aokp is true. See memoize-use-attachment-warning for how
; this special lookup-marker (:lookup . fn) is used.
`((update-attached-fn-called ',(cons :lookup fn))))
,@mf-record-hit
,@(cond ((null (cdr stobjs-out))
`(,*mf-ans*))
(t
; The result stored in the memo table is of the form (list* x1 x2 ... xk),
; where k is the length of stobjs-out. This result was stored when the
; returned value was (mv x1 x2 ... xk). We build an expression (mv e1 e2
; ... ek), where each ei except the last is the result of popping the variable
; stored in *mf-ans*, i.e., #:ANS, leaving ek as that variable itself.
(let ((len-1 (1- (length stobjs-out))))
`(,(cons
'mv
(nconc (loop for i fixnum below len-1
collect `(pop ,*mf-ans*))
(list *mf-ans*))))))))
(t
; Memoized lookup failed. Now call the function and perhaps store what is
; returned by the call.
,(let* ((vars
; Vars is the list of variables (O0 O1 ... 0N) for the outputs, where N is the
; length of stobjs-out. (We play it safe and use (O0) if somehow stobjs-out is
; nil.)
(loop for i fixnum below
(if (cdr stobjs-out)
(length stobjs-out)
1)
collect (mf-make-symbol "O~a" i))))
(cond
(aokp
`(mv?-let
,vars
,body
(progn (mf-sethash ,*mf-args*
(list* ,@vars)
,localtablename)
(mv? ,@vars))))
(t ; (not aokp); so avoid using attachments
`(let (,*attached-fn-temp*)
(mv?-let
,vars
(let ((*aokp* (and *aokp* t)))
(,prog1-fn
,body
(when (not (member *aokp* '(t nil) :test #'eq))
(setq ,*attached-fn-temp* *aokp*))))
(progn
(cond
(,*attached-fn-temp*
(memoize-use-attachment-warning
',fn
,*attached-fn-temp*))
(t ; Save the results in the memo table.
(mf-sethash ,*mf-args*
(list* ,@vars)
,localtablename)))
; Special variable *aokp* was protected by a LET above when executing body,
; which could have set it to a function symbol whose attachment was used. That
; value was however saved in *attached-fn-temp*, which we now use to update
; *aokp* since *sokp* was restored, after the execution of body, to its
; previous value (by popping out of its binding scope).
(when ,*attached-fn-temp* ; optimization
(assert *aokp*)
(update-attached-fn-called ,*attached-fn-temp*))
(mv? ,@vars))))))))))))))))
(defun memoize-fn-outer-body (inner-body fn fn-col-base start-ticks forget
number-of-args
tablename ponstablename
prog1-fn)
; The code returned below will be put in the context of outer-body in the code
; returned by memoize-fn-def. It is a wrapper for inner-body, and it is used
; when the special variable symbol stored in start-ticks has a symbol-value of
; -1, indicating that we are looking at an outermost call of fn. This
; mechanism avoids double-counting for recursive calls of fn.
; Comments about "performance counting" mark code that might best be ignored on
; a first reading. A long comment in *memoize-call-array* outlines how
; performance counting is implemented.
; This code is what sets the special variable, *caller*, to the appropriate
; index for fn into *memoize-call-array* (namey fn-col-base, which points to
; the column for fn). So, any immediately subsidiary calls of memoized
; functions will be "charged" to fn, and *caller* will only be re-assigned by
; subsidiary top-level calls of other memoized functions.
;; [Jared] multithreading note: all of this is going to be protected with the
;; global lock, so we shouldn't need any locking inside here. See
;; memoize-fn-def.
`(let (#+(or ccl sbcl)
,@(and *record-bytes* ; performance counting
`((,*mf-start-bytes* (heap-bytes-allocated))))
,@(and *record-pons-calls* ; performance counting
`((,*mf-start-pons* *pons-call-counter*)))
(,start-ticks ,(if *record-time*
'(internal-real-ticks)
'0)))
(declare
(ignorable #+(or ccl sbcl)
,@(and *record-bytes* `(,*mf-start-bytes*))
,@(and *record-pons-calls* `(,*mf-start-pons*)))
(type mfixnum
,start-ticks
,@(and *record-pons-calls* `(,*mf-start-pons*))
#+(or ccl sbcl)
,@(and *record-bytes* `(,*mf-start-bytes*))))
(,(cond ((or *record-pons-calls*
*record-bytes*
*record-time*
forget)
; The test above is the condition under which there are forms following the
; inner-body call below. If there are no such forms, we simply lay down a
; progn (which presumably will be compiled away).
prog1-fn)
(t 'progn))
; At one time we used unwind-protect here (actually, we used
; unwind-protect-disable-interrupts-during-cleanup). But some very simple
; experiments show that prog1 and multiple-value-prog1 are much cheaper than
; unwind-protect, even if we introduce *caller* and start-ticks as special
; variables (so that they can be let-bound) instead of defg. Maybe that's in
; the noise, but it seems worth trying, even if when aborting computations,
; some of the statistics are skewed or, if forget is true, some tables fail to
; be flushed.
,(cond ((or *record-bytes*
*record-calls*
*record-hits*
*record-mht-calls*
*record-pons-calls*
*record-time*)
; Then we are gathering performance counting statistics for this function, so
; we need to make a note of the caller. Otherwise, this function is invisible
; from the standpoint of performance counting, and we should let the caller
; stay as it was.
`(let ((*caller* ,fn-col-base)) ; performance counting
,inner-body))
(t inner-body))
,@(and *record-pons-calls* ; performance counting
`((safe-incf
(aref ,*mf-ma*
,(ma-index-from-col-base fn-col-base *ma-pons-index*))
(the-mfixnum (- *pons-call-counter* ,*mf-start-pons*)))))
#+(or ccl sbcl)
,@(and *record-bytes* ; performance counting
`((safe-incf
(aref ,*mf-ma*
,(ma-index-from-col-base fn-col-base *ma-bytes-index*))
(the-mfixnum (- (heap-bytes-allocated) ,*mf-start-bytes*)))))
,@(and *record-time* ; performance counting
`((safe-incf
(the mfixnum
(aref ,*mf-ma* (the-mfixnum (1+ ,*mf-count-loc*))))
(the-mfixnum
(fix-ticks (the-mfixnum (- (internal-real-ticks)
,start-ticks))
',fn)))))
,@(and forget
; Below, we "clear" the tables by setting them to nil, rather than by calling
; clrhash. The latter approach would probably avoid reallocation of space, but
; we suspect that such a gain in space efficiency might be offset by a loss in
; time efficiency. The present approach has been working, so we prefer not to
; change it. Below is just a little space analysis.
; Perhaps we should use clrhash below. Here is the story:
; When we evaluated (defconstant *a* (make-list 1000)) in raw Lisp (CCL) and
; then ran (time$ (bad-lisp-objectp *a*)), we saw about 70K bytes allocated
; then and each additional time (time$ (bad-lisp-objectp *a*)) was evaluated,
; because the :forget argument (see *thread-unsafe-builtin-memoizations*, used
; in acl2h-init-memoizations) was blowing away hash tables -- see setq forms
; below. After evaluating (unmemoize-fn 'bad-lisp-objectp), the bytes went to
; 0 per evaluation. Then we evaluated (memoize-fn 'bad-lisp-objectp) -- this
; time without a :forget argument -- and found about 70K bytes allocated on the
; first timing of (bad-lisp-objectp *a*), but 0 bytes allocated on subsequent
; evaluations, presumably because we weren't starting from scratch.
; Interestingly, the byte allocation on subsequent runs was also 0 after
; memoizing with :forget t when we experimented by replacing the setq forms
; below by clrhash calls. So perhaps we should consider using clrhash instead
; of setq below, to avoid generating garbage hash tables.
`((setq ,tablename nil)
,@(and (> number-of-args 1)
`((setq ,ponstablename nil))))))))
(defun memoize-fn-def (inner-body outer-body
fn formals specials dcls fnn start-ticks
localtablename localponstablename)
; Comments about "performance counting" mark code that might best be ignored on
; a first reading. A long comment in *memoize-call-array* outlines how
; performance counting is implemented.
; See example below.
`(defun ,fn ,formals ,@dcls
(declare (ignorable ,@formals ,@specials))
(with-global-memoize-lock-static
;; Special case. This is the lock that will be grabbed when executing
;; the memoized function. For maximum performance in the single threaded
;; case, we use the less-safe, static version of lock grabbing.
(let* ((,*mf-count-loc* ; performance counting
,(if (or *record-calls* *record-time*)
; The following expression is similar to `(ma-index-calls ,fnn *caller*),
; except that *caller* is already a column (that is, already multiplied by the
; square dimension of *memoize-call-array*).
`(the-mfixnum (+ *caller* ,(* 2 fnn)))
0))
(,*mf-ma* *memoize-call-array*)
,localtablename ,localponstablename)
(declare (type mfixnum ,*mf-count-loc*)
(ignorable ,*mf-count-loc* ,*mf-ma*
,localponstablename
,localtablename)
(type (simple-array mfixnum (*))
,*mf-ma*))
,@(and *record-calls* ; performance counting
`((safe-incf (aref ,*mf-ma* ,*mf-count-loc*) 1)))
; At one time we introduced an flet to share code for the original body, which
; can occur twice or three times below. But we changed this on 10/18/2014
; because we think that the code expansion might be worth it, in case flet
; reduces performance. Probably this is not a sufficiently important issue to
; pursue further.
(if (eql -1 ,start-ticks)
,outer-body
,inner-body)))))
; Here is an example of the memoization of a simple function, fib, defined as:
; ACL2 !>(defun fib (n)
; (declare (xargs :guard (natp n)))
; (if (zp n) 0 (if (eql n 1) 1 (+ (fib (- n 1)) (fib (- n 2))))))
; In the following, let the two long names in the ACL2_INVISIBLE package
; below be denoted by the short names shown:
; long name short name
; ACL2_INVISIBLE::|HONS-G-"\\"START-TICKS-\\",\\"ACL2\\",\\"FIB\\"",13| sticks
; ACL2_INVISIBLE::|HONS-G-"\\"MEMOIZE-HT-FOR-\\",\\"ACL2\\",\\"FIB\\"",14| mht
; Here is the raw Lisp defun of fib produced by (memoize 'fib). It was
; obtained as follows.
; :q
; (old-trace memoize-fn-def)
; (lp)
; (memoize 'fib)
; ; Then copy result of memoize-fn-def into a file foo, make the two
; ; replacements above (globally), and save foo after:
; ; (setq xxx (defun ...) )
; :q
; (load foo)
; xxx
; (pprint *)
; ; Now replace STICKS and MHT by sticks and mht, resp., in the output.
; (DEFUN FIB (N)
; (DECLARE (XARGS :GUARD (NATP N)))
; (DECLARE (IGNORABLE N))
; (WITH-GLOBAL-MEMOIZE-LOCK-STATIC
; (LET* ((#:MF-COUNT-LOC (THE-MFIXNUM (+ *CALLER* 18)))
; (#:MA *MEMOIZE-CALL-ARRAY*)
; #:TABLENAME
; #:PONSTABLENAME)
; (DECLARE (TYPE MFIXNUM #:MF-COUNT-LOC)
; (IGNORABLE #:MF-COUNT-LOC #:MA #:PONSTABLENAME #:TABLENAME)
; (TYPE (SIMPLE-ARRAY MFIXNUM (*)) #:MA))
; (SAFE-INCF (AREF #:MA #:MF-COUNT-LOC) 1)
; (IF (EQL -1 sticks)
; (LET ((#:START-BYTES (HEAP-BYTES-ALLOCATED))
; (#:START-PONS *PONS-CALL-COUNTER*)
; (sticks (INTERNAL-REAL-TICKS)))
; (DECLARE (IGNORABLE #:START-BYTES #:START-PONS)
; (TYPE MFIXNUM sticks #:START-PONS #:START-BYTES))
; (PROG1 (LET ((*CALLER* 9000))
; (COND (T
; (LET (#:ANS #:ARGS #:ANS-P)
; (DECLARE
; (IGNORABLE #:ANS #:ARGS #:ANS-P))
; (WHEN
; (NULL mht)
; (SAFE-INCF (AREF #:MA 9002) 1)
; (SETQ
; mht
; (MAKE-INITIAL-MEMOIZE-HASH-TABLE
; 'FIB
; 60)))
; (SETQ #:TABLENAME mht)
; (SETQ #:ARGS (PIST* #:PONSTABLENAME N))
; (MULTIPLE-VALUE-SETQ
; (#:ANS #:ANS-P)
; (MF-GETHASH #:ARGS #:TABLENAME))
; (COND
; (#:ANS-P
; (SAFE-INCF (AREF #:MA 9001) 1)
; #:ANS)
; (T
; (LET
; (#:ATTACHED-FN-TEMP)
; (MV?-LET
; (O0)
; (LET
; ((*AOKP* (AND *AOKP* T)))
; (PROG1
; (IF
; (ZP N)
; 0
; (IF
; (EQL N 1)
; 1
; (+
; (FIB (- N 1))
; (FIB (- N 2)))))
; (WHEN
; (NOT
; (MEMBER
; *AOKP*
; '(T NIL)
; :TEST
; #'EQ))
; (SETQ
; #:ATTACHED-FN-TEMP
; *AOKP*))))
; (PROGN
; (COND
; (#:ATTACHED-FN-TEMP
; (MEMOIZE-USE-ATTACHMENT-WARNING
; 'FIB
; #:ATTACHED-FN-TEMP))
; (T
; (MF-SETHASH
; #:ARGS
; (LIST* O0)
; #:TABLENAME)))
; (WHEN
; #:ATTACHED-FN-TEMP
; (ASSERT *AOKP*)
; (UPDATE-ATTACHED-FN-CALLED
; #:ATTACHED-FN-TEMP))
; (MV? O0))))))))))
; (SAFE-INCF (AREF #:MA 9004)
; (THE-MFIXNUM
; (- *PONS-CALL-COUNTER* #:START-PONS)))
; (SAFE-INCF (AREF #:MA 9000)
; (THE-MFIXNUM
; (-
; (HEAP-BYTES-ALLOCATED)
; #:START-BYTES)))
; (SAFE-INCF (THE
; MFIXNUM
; (AREF
; #:MA
; (THE-MFIXNUM (1+ #:MF-COUNT-LOC))))
; (THE-MFIXNUM
; (FIX-TICKS
; (THE-MFIXNUM
; (- (INTERNAL-REAL-TICKS) sticks))
; 'FIB)))))
; (COND (T
; (LET (#:ANS #:ARGS #:ANS-P)
; (DECLARE (IGNORABLE #:ANS #:ARGS #:ANS-P))
; (WHEN (NULL mht)
; (SAFE-INCF (AREF #:MA 9002) 1)
; (SETQ mht
; (MAKE-INITIAL-MEMOIZE-HASH-TABLE
; 'FIB
; 60)))
; (SETQ #:TABLENAME mht)
; (SETQ #:ARGS (PIST* #:PONSTABLENAME N))
; (MULTIPLE-VALUE-SETQ (#:ANS #:ANS-P)
; (MF-GETHASH #:ARGS #:TABLENAME))
; (COND (#:ANS-P
; (SAFE-INCF (AREF #:MA 9001) 1)
; #:ANS)
; (T
; (LET (#:ATTACHED-FN-TEMP)
; (MV?-LET (O0)
; (LET
; ((*AOKP* (AND *AOKP* T)))
; (PROG1
; (IF
; (ZP N)
; 0
; (IF
; (EQL N 1)
; 1
; (+
; (FIB (- N 1))
; (FIB (- N 2)))))
; (WHEN
; (NOT
; (MEMBER
; *AOKP*
; '(T NIL)
; :TEST
; #'EQ))
; (SETQ
; #:ATTACHED-FN-TEMP
; *AOKP*))))
; (PROGN
; (COND
; (#:ATTACHED-FN-TEMP
; (MEMOIZE-USE-ATTACHMENT-WARNING
; 'FIB
; #:ATTACHED-FN-TEMP))
; (T
; (MF-SETHASH
; #:ARGS
; (LIST* O0)
; #:TABLENAME)))
; (WHEN
; #:ATTACHED-FN-TEMP
; (ASSERT *AOKP*)
; (UPDATE-ATTACHED-FN-CALLED
; #:ATTACHED-FN-TEMP))
; (MV? O0)))))))))))))
(defun-one-output memoize-eval-compile (def old-fn)
#+(or ccl sbcl) ; all functions are compiled
(declare (ignore old-fn))
(unless (and (consp def)
(eq 'defun (car def))
(consp (cdr def))
(symbolp (cadr def)))
(error "MEMOIZE-EVAL-COMPILE: Bad input:~%~s." def))
#+(or ccl sbcl)
(compile (eval def)) ; probably the compile is unnecessary but harmless
#-(or ccl sbcl)
(if (and old-fn (compiled-function-p old-fn))
(compile (eval def))
(eval def))
nil)
(defun memoize-fn-formals (fn wrld inline
&key
(cl-defun :default cl-defun-p))
; This function is used not only below, but also in community book
; books/centaur/memoize/old/profile.lisp (actually profile-raw.lsp).
(let ((formals (getprop fn 'formals t 'current-acl2-world wrld)))
(if (eq formals t)
(let ((cl-defun (if cl-defun-p
cl-defun
(memoize-look-up-def fn cl-defun inline wrld
:quiet-p t))))
(if (consp cl-defun)
(cadr cl-defun)
(let ((n (mf-len-inputs fn)))
(if n
(loop for i fixnum below n
collect (mf-make-symbol "X~a" i))
t ; error
))))
formals)))
(defun memoize-fn (fn &key
(condition t)
(inline t)
(cl-defun :default)
(formals :default)
(stobjs-in :default)
(specials nil)
(commutative nil)
(stobjs-out :default)
(forget nil)
(memo-table-init-size *mht-default-size*)
(aokp nil)
&aux
(wrld (w *the-live-state*))
(memoize-fn-signature-error
"Memoize-fn: could not determine a signature for ~a.~%~
To assert the signature of ~:*~a, use (mf-note-arity '~:*~a nargs nvals),~%~
for example, (mf-note-arity '~:*~a 3 1).) If there are stobjs arguments~%~
then do not use that mechanism; instead, pass suitable arguments to~%~
memoize-fn."))
; Warning: Keep the argument list for this function in sync with memoize-info,
; as explained there.
; We do not claim that memoize-fn is bulletproof when called directly in raw
; Lisp, rather than via a call to memoize in the ACL2 loop. However, we make
; some effort for it to behave reasonably.
; If the :CONDITION parameter is NIL we record performance data but not the
; values computed; this is used in our profiling tool. Indeed, simple
; profiling is just memoizing with :condition nil and :inline nil.
; If the :INLINE parameter is T, then the body of FN is placed inline in the
; memoized definition; otherwise, a funcall of the original function is placed
; there.
; Comments about "performance counting" mark code that might best be ignored on
; a first reading. A long comment in the Essay on the *Memoize-Call-Array*
; outlines how performance counting is implemented.
; Old documentation, which may be updated soon and/or merged into the :doc:
"The documentation for MEMOIZE-FN is very incomplete. One may invoke
(MEMOIZE-FN fn) on the name of a Common Lisp function FN from outside the
ACL2 loop to get results of possible interest in terms of memoization
activity and profiling information. MEMOIZE-FN already has a dozen
parameters.
MEMOIZE-FN replaces the SYMBOL-FUNCTION for the symmbol FN with 'enhanced'
raw Common Lisp code that, supposedly, does not affect the value returned by
FN but may make some notes and may even obtain some return values by
remembering them from a previous call.
If the CONDITION parameter is not NIL or 'NIL, then whenever FN is called,
and there is not as yet any value remembered for a call of FN on the given
arguments, then if the evaluation of the CONDITION parameter (where the
formals of FN are bound to its actuals) is not NIL or 'NIL, the values that
are computed for FN on the given arguments will be saved under the given
arguments.
If the INLINE parameter is T, then when MEMOIZE-FN creates a new body for FN,
we place the old body of FN within the new body, i.e., 'in line'. However,
if the INLINE parameter is NIL, then we place code that calls the existing
SYMBOL-FUNCTION for FN within the new body. One might well argue that our
parity for the INLINE parameter to MEMOIZE-fn is backwards, but we don't
think so.
One may lie to MEMOIZE-FN to force the memoization of a function that has
ACL2's state as an explicit parameter by using fraudulent FORMALS, STOBJS-IN,
and STOBJS-OUT parameters to MEMOIZE-FN.
If the COMMUTATIVE parameter is not NIL, then the two arguments may be
swapped before further processing. We hope/presume that ACL2 will have been
used first to prove that commutativity.
If the CL-DEFUN parameter is not NIL, we pretend that the current body of FN
is that parameter, and similarly for FORMALS, STOBJS-IN, and STOBJS-OUT.
If FN is a raw Common Lisp function and not an ACL2-approved function, it may
make reference to a variable, say S, that has a SPECIAL binding, in which
case one needs to consider what in the world the meaning is for memoizing
such a function at all. If S is a member of the SPECIALS parameter, then it
is assumed that FN does not alter but only refers to S. MEMOIZE-FN acts as
though FN had S as an extra argument for purposes of memoization.
If the FORGET parameter is not NIL, the pons and memo tables of FN are
discarded at the end of every outermost call of FN."
#-hons
(return-from memoize-fn
(progn (when (not (zerop *ld-level*))
(warning$ 'memoize nil
"No change for function ~x0: Memoization ~
requests are ignored in this ACL2 ~
executable because it is not hons-enabled."
fn))
fn))
(when (equal condition *nil*)
(setq condition nil))
(with-global-memoize-lock
;; [Jared] note that this lock protects the operation of the memoize
;; command, itself. The definition of the newly memoized function has
;; its own, separate lock. See memoize-fn-def.
; Perhaps this lock is not necessary; after all, it seems unlikely that
; different threads would be memoizing at the same time. On the other hand, we
; really don't want to worry about the possibility that some structures are
; installed and others are not if another thread tries to call a function that
; is in the process of being memoized; so we use a lock here.
(with-warnings-suppressed ; e.g., might avoid redefinition warnings
(memoize-fn-init fn inline wrld)
(let* ((cl-defun (memoize-look-up-def fn cl-defun inline wrld))
(formals
(cond
((eq formals :default)
(let ((formals (memoize-fn-formals fn wrld inline
:cl-defun cl-defun)))
(if (eq formals t)
(error memoize-fn-signature-error fn)
formals)))
(t formals)))
(stobjs-in
(if (eq stobjs-in :default)
(let ((s (getprop fn 'stobjs-in t 'current-acl2-world wrld)))
(if (eq s t)
(make-list (len formals)) ; assume no stobjs
s))
stobjs-in))
(stobjs-out
(if (eq stobjs-out :default)
(let ((s
; Note that the never-memoize-p check made in memoize-fn-init guarantees that
; here, we are not taking the stobjs-out of functions in *stobjs-out-invalid*.
(getprop fn 'stobjs-out t 'current-acl2-world wrld)))
(if (eq s t)
(let ((n (mf-len-outputs fn)))
(cond (n (make-list n))
((null condition)
; This case is impossible when memoization is invoked using memoize inside the
; ACL2 loop, so we do not need to expect perfection here. Fortunately,
; stobjs-out is probably irrelevant when condition is nil; so we assume that a
; single non-stobj value is returned.
(list nil))
(t (error memoize-fn-signature-error fn))))
s))
stobjs-out)))
(memoize-fn-init-2 fn formals condition stobjs-in stobjs-out)
(mf-sethash fn
(cons (length stobjs-in) (length stobjs-out))
*number-of-arguments-and-values-ht*)
(let* ((fnn ; performance counting
(symbol-to-fixnum-create fn))
(fn-col-base ; performance counting
(ma-index fnn))
(old-fn (symbol-function fn))
(body
(progn
(assert old-fn)
(cond
(inline
; When memoize-look-up-def is called with a non-nil value of inline and it
; returns without error, then it returns a non-nil definition (which is a true
; list).
(assert (consp cl-defun))
(car (last cl-defun)))
(t `(funcall
#-gcl ,old-fn
#+gcl ',old-fn ; old-fn could be (lisp:lambda-block ...)
,@formals)))))
(dcls (mf-dcls (cddr (butlast cl-defun))))
(start-ticks
; Submit a defg form and return the generated name:
(let ((v (hons-gentemp
(memoize-fn-suffix "START-TICKS-" fn))))
(eval `(prog1 (defparameter ,v -1)
(declaim (type mfixnum ,v))))))
(tablename
; Submit a defg form and return the generated name:
(eval `(defg
,(hons-gentemp (memoize-fn-suffix "MEMOIZE-HT-FOR-" fn))
nil)))
(ponstablename
; Submit a defg form and returns the generated name:
(eval `(defg
,(hons-gentemp (memoize-fn-suffix "PONS-HT-FOR-" fn))
nil)))
(localtablename (make-symbol "TABLENAME"))
(localponstablename (make-symbol "PONSTABLENAME"))
(sts
; Here we support memoize-flush, which keeps memoize tables coherent in the
; presence of user-defined stobjs. For each (user-level) stobj input name, st,
; we collect up the variable (st-lst st), whose value is the list of names of
; memoize tables that need to be cleared whenever that stobj changes. Below,
; we will push the present function's table name onto each of these lists.
; Ultimately, stobj updates are made by stobj primitives. After the stobj
; primitive for stobj st makes an update, the memo table for a function f
; taking st as its nth argument (say) is no longer valid for saved calls of f
; for which the nth argument is st. Because of congruent stobjs and abstract
; stobjs, that nth argument need not be st, and we believe that in principle,
; we could restrict flushing of memo table entries of f to those whose nth
; argument is eq to the stobj being updated (whether st, congruent to st, or an
; abstract stobj whose concrete stobj is congruent to st). But for now we take
; the coarser approach, which has the advantage that we simply throw away the
; memo-table for f when flushing, leaving it to be garbage collected; see
; memoize-flush1.
(and condition ; else no memo table usage, so skip flushing
(remove-duplicates-eq
(loop for st in (union stobjs-in stobjs-out)
when st
collect
(assert$
(not (eq st 'state)) ; see memoize-table-chk
(st-lst (congruent-stobj-rep
; In the case that st is an abstract stobj, we replace it with the
; corresponding concrete stobj before getting a canonical (congruent)
; representative; see the rather long comment just above that mentions
; "abstract" stobjs.
(or (concrete-stobj st wrld)
st)
wrld)))))))
(number-of-args
; This is the number of arguments. Specials only matter for raw Lisp
; functions; see the notes above in memoize-fn. Basically if the function
; reads from specials we want to count them as arguments.
(+ (len formals) (len specials)))
(prog1-fn (if (cdr stobjs-out)
'multiple-value-prog1
'prog1))
(inner-body
(memoize-fn-inner-body
fn condition body fn-col-base
tablename localtablename ponstablename localponstablename
memo-table-init-size
formals specials stobjs-out number-of-args
commutative aokp prog1-fn))
(outer-body
(memoize-fn-outer-body
inner-body fn fn-col-base start-ticks forget number-of-args
tablename ponstablename prog1-fn))
(def
(memoize-fn-def
inner-body outer-body
fn formals specials dcls fnn start-ticks
localtablename localponstablename))
(success nil))
(unwind-protect
(progn
(let ((ma *memoize-call-array*))
(declare (type (simple-array mfixnum (*)) ma))
(loop for i of-type mfixnum ; performance counting
from fn-col-base ; loop through this column
below
(ma-index-from-col-base fn-col-base *2max-memoize-fns*)
unless (eql (aref ma i) 0)
do (setf (aref ma i) 0)))
(memoize-eval-compile def old-fn)
(mf-sethash fn
(make memoize-info-ht-entry
:ext-anc-attachments
(and aokp (ext-ancestors-attachments fn wrld))
:start-ticks start-ticks
:num fnn
:tablename tablename
:ponstablename ponstablename
:condition condition
:inline inline
:memoized-fn (symbol-function fn)
:old-fn old-fn
:fn fn
:sts sts
:cl-defun cl-defun
:formals formals
:commutative commutative
:specials specials
:stobjs-in stobjs-in
:stobjs-out stobjs-out
:record-bytes *record-bytes*
:record-calls *record-calls*
:record-hits *record-hits*
:record-mht-calls *record-mht-calls*
:record-pons-calls *record-pons-calls*
:record-time *record-time*
:forget forget
:memo-table-init-size memo-table-init-size
:aokp aokp)
*memoize-info-ht*)
(mf-sethash fnn fn *memoize-info-ht*)
(and condition (loop for s in sts do
(push tablename
(symbol-value (the symbol s)))))
(setq success t))
(unless success
(without-interrupts ; perhaps already the case in some Lisps
(setf (symbol-function fn) old-fn)
(remhash fn *memoize-info-ht*)
(remhash fnn *memoize-info-ht*)
(and condition
(loop for s in sts
when (eq tablename
(car (symbol-value (the symbol s))))
do (pop (symbol-value (the symbol s)))))
(setq fn nil) ; in case we decide not to cause an error
(error "Memoize-fn: Failed to memoize ~s." fn))))))))
fn)
(defun-one-output unmemoize-fn (fn)
; Warning: Do not be tempted to remove the definition of the condition
; function. If fn was memoized using the memoize macro in the ACL2 loop, then
; a non-trivial condition is a first-class ACL2 function.
#-hons
(return-from unmemoize-fn
(progn (when (not (zerop *ld-level*))
(warning$ 'unmemoize nil
"No change for function ~x0: Unmemoization ~
requests are ignored in this ACL2 ~
executable because it is not hons-enabled."
fn))
fn))
(with-global-memoize-lock
(maybe-untrace! fn) ; See the comment about Memoization in trace$-def.
(let ((rec (memoizedp-raw fn)))
(unless rec ; expect rec=nil in the loop, else unmemoize is redundant
(error "Unmemoize-fn: ~s is not memoized." fn))
(let* ((num (the-mfixnum (access memoize-info-ht-entry rec
:num)))
(tablename (access memoize-info-ht-entry rec
:tablename))
(ponstablename (access memoize-info-ht-entry rec
:ponstablename))
(sts (access memoize-info-ht-entry rec
:sts))
(col-base (ma-index num))
(saved-table (symbol-value (the symbol tablename)))
(saved-ponstable (symbol-value (the symbol ponstablename)))
(saved-fn-entry (mf-gethash fn *memoize-info-ht*))
(saved-num-entry (mf-gethash num *memoize-info-ht*))
(ma *memoize-call-array*)
(success nil))
(declare (type mfixnum num col-base)
(type (simple-array mfixnum (*)) ma))
(unwind-protect
(progn
(format nil "unmemoizing ~s" fn)
(let (#+ccl (ccl:*warn-if-redefine-kernel* nil))
(let ((old-fn (access memoize-info-ht-entry rec :old-fn)))
(assert old-fn)
(setf (symbol-function (the symbol fn))
old-fn)))
(loop for i of-type mfixnum
from col-base ; loop through col-base column
below (ma-index-from-col-base col-base *2max-memoize-fns*)
unless (eql (aref ma i) 0)
do (setf (aref ma i) 0))
(remhash fn *memoize-info-ht*)
(remhash num *memoize-info-ht*)
(setf (symbol-value (the symbol tablename)) nil)
(setf (symbol-value (the symbol ponstablename)) nil)
(loop for s in sts do
(when (boundp s)
(setf (symbol-value (the symbol s))
(remove tablename (symbol-value (the symbol s))))))
(setq success t))
(unless success
(without-interrupts ; perhaps already the case in some Lisps
; We don't restore *memoize-call-array*; those statistics are simply lost.
(setf (symbol-value (the symbol tablename)) saved-table)
(setf (symbol-value (the symbol ponstablename)) saved-ponstable)
(mf-sethash fn saved-fn-entry *memoize-info-ht*)
(mf-sethash num saved-num-entry *memoize-info-ht*)
(loop for s in sts do
(when (boundp s)
(push tablename (symbol-value (the symbol s)))))))))))
fn)
(defun-one-output maybe-unmemoize (fn)
; We rely on the normal undoing mechanism (for :ubt etc.) to take care of
; unmemoization. However, as a courtesy to users who memoize using raw Lisp,
; we provide this interface for unmemoizing functions that might not be known
; to ACL2 (via the memoize-table) as being memoized.
(with-global-memoize-lock
; We use a lock here to prevent the possibility that another thread unmemoizes
; after our memoizedp-raw test but before our unmemoize-fn call. We don't
; expect many calls of maybe-unmemoize, so this lock should be cheap.
(when (and (memoizedp-raw fn)
(not (cdr (assoc-eq fn (table-alist 'memoize-table
(w *the-live-state*))))))
(unmemoize-fn fn))))
(defun-one-output memoized-functions ()
; We return the names of all memoized functions as a list.
(with-global-memoize-lock
(let (ans)
(mf-maphash (lambda (fn info)
(declare (ignore info))
(when (symbolp fn)
(push fn ans)))
*memoize-info-ht*)
ans)))
(defun-one-output length-memoized-functions ()
; We return the number of currently memoized (or profiled) functions. We
; subtract one for the outside-caller entry and divide by 2 since there are two
; entries for each memoized function: one for the function symbol and one that
; associates :num of that symbol's entry back to the symbol.
(with-global-memoize-lock
(values (floor (1- (mf-hash-table-count *memoize-info-ht*))
2))))
(defun-one-output unmemoize-all ()
; Unmemoize all currently memoized functions,including all profiled functions.
; Warning: This function does not deal with memoize-table. ACL2 users should
; evaluate (clear-memo-table) instead of calling unmemoize-all directly.
; A warning to would-be code improvers. It would be a bad idea to redefine
; unmemoize-all to maphash over *memoize-info-ht*, because of the ANSI rules
; restricting which hash table entries may be modified during a maphash.
(loop for x in (memoized-functions) do (unmemoize-fn x)))
(defun-one-output memoize-info (k)
; Warning: Keep this in sync with memoize-fn and rememoize-all.
; We return the settings of the various arguments to memoize-fn and the
; settings of the special variables that influence memoize-fn during the
; memoization of the given function symbol. The rather arcane structure
; returned is used in rememoize-all.
(with-global-memoize-lock
(let ((v (mf-gethash k *memoize-info-ht*)))
(and v
(list
; Warning: Keep the following list in sync with the argument list for
; memoize-fn. Also note that the :fn must be first; see rememoize-all.
(list (access memoize-info-ht-entry v :fn)
:condition
(access memoize-info-ht-entry v :condition)
:inline
(access memoize-info-ht-entry v :inline)
:cl-defun
(access memoize-info-ht-entry v :cl-defun)
:formals
(access memoize-info-ht-entry v :formals)
:stobjs-in
(access memoize-info-ht-entry v :stobjs-in)
:specials
(access memoize-info-ht-entry v :specials)
:commutative
(access memoize-info-ht-entry v :commutative)
:stobjs-out
(access memoize-info-ht-entry v :stobjs-out)
:forget
(access memoize-info-ht-entry v :forget)
:memo-table-init-size
(access memoize-info-ht-entry v :memo-table-init-size)
:aokp
(access memoize-info-ht-entry v :aokp))
; Warning: Keep the following list in sync with the set of *record-xxx*
; globals, and with rememoize-all.
(list (access memoize-info-ht-entry v :record-bytes)
(access memoize-info-ht-entry v :record-calls)
(access memoize-info-ht-entry v :record-hits)
(access memoize-info-ht-entry v :record-mht-calls)
(access memoize-info-ht-entry v :record-pons-calls)
(access memoize-info-ht-entry v :record-time)))))))
(defun-one-output rememoize-all ()
; Warning: Keep this function in sync with memoize-info.
(with-global-memoize-lock
(let (lst)
(mf-maphash (lambda (k v)
(declare (ignore v))
(when (symbolp k)
(push (memoize-info k) lst)))
*memoize-info-ht*)
; Note: memoize-info arranges that (caar x) is the memoized function symbol.
(loop for x in lst do (unmemoize-fn (caar x)))
(setq *memoize-info-ht* (initial-memoize-info-ht))
(gc$)
(setq *max-symbol-to-fixnum* *initial-max-symbol-to-fixnum*)
(loop for x in lst do
; Warning: Keep the first argument below in sync with memoize-info.
(progv '(*record-bytes*
*record-calls*
*record-hits*
*record-mht-calls*
*record-pons-calls*
*record-time*)
(cadr x)
(apply 'memoize-fn (car x)))))))
(defun profile-fn (fn &rest r &key (condition nil) (inline nil)
&allow-other-keys)
; Note that :condition and :inline are laid down twice in the call of
; memoize-fn below: once explicitly and once in r. There is no error in such
; repetition, as Section 3.4.1.4 ("Specifiers for keyword parameters") of the
; Common Lisp Hyperspec says: "If more than one such argument pair matches, the
; leftmost argument pair is used." By including the :condition and :inline
; arguments explicitly below, which are specified in the argument list as nil,
; we guarantee that the resulting function is profiled and executes its
; original body.
(apply #'memoize-fn fn
:condition condition
:inline inline
r))
(defun-one-output profiled-functions ()
; For purposes of this function, the profiled functions are defined as those
; produced by memoize-fn with null :condition and :inline fields. We could
; probably allow :inline to be non-nil, but it will always be nil for functions
; profiled using profile-fn in raw Lisp or profile in the loop, so we enforce a
; nil value for :inline, not just :condition.
(with-global-memoize-lock
(let (lst)
(mf-maphash
(lambda (k v)
(when (and (symbolp k)
(null (access memoize-info-ht-entry v :condition))
(null (access memoize-info-ht-entry v :inline)))
(push k lst)))
*memoize-info-ht*)
lst)))
(defun-one-output unmemoize-profiled ()
; This raw Lisp function unmemoizes (and hence unprofiles) all functions
; currently memoized with :condition nil and :inline nil.
(loop for x in (profiled-functions) do
(unmemoize-fn (car x))))
; SECTION: Statistics gathering and printing routines
(defun-one-output coerce-index (x)
; X is either a memoized function symbol or the associated number
; (symbol-to-fixnum, also column number in *memoize-call-array* and index in
; *callers-array*) of such.
; Note: We might be tempted to lock uses of this function, since it calls
; symbol-to-fixnum, which uses a global hash-table. But currently that is not
; necessary: all callers of coerce-index are functions defined in this Section,
; and their intended use is in sorting of memsum results; see
; *memoize-summary-order-list*.
(cond ((typep x 'fixnum) ; equivalently, x is a number
(assert (and (>= (the fixnum x) 0)
(< (the fixnum x)
(the fixnum (ash (length *memoize-call-array*)
-1)))))
x)
(t ; could (assert (symbolp x)) but probably not worth the cycles
(symbol-to-fixnum x))))
(defun-one-output mf-num-to-string (n) ; for forming numbers
; Consider invoking this function in the context of our-syntax.
(check-type n number)
(if (= n 0) (setq n 0))
(cond ((typep n '(integer -99 999))
(format nil "~8d" n))
((or (< -1000 n -1/100)
(< 1/100 n 1000))
(format nil "~8,2f" n))
(t (format nil "~8,2e" n))))
(defun mf-shorten (x n)
; Consider invoking this function in the context of our-syntax.
; Unless x is a string of length at most n, return a string that represents x,
; truncated to the first n characters, and associate x with that string in
; *mf-shorten-ht* (if there is not already such association).
(cond ((and (stringp x) (<= (length x) n))
x)
(t
(let ((x (if (stringp x)
(hons-copy x)
x)))
;; [Jared] I think there's no need to protect this hash table
;; because of the way it's used.
(cond ((mf-gethash x *mf-shorten-ht*))
(t (let ((*print-pretty* nil)
(*print-length* 10)
(*print-level* 5)
(*print-lines* 2))
(let ((str
(with-output-to-string
(s)
(cond ((stringp x) (princ x s))
(t (prin1 x s))))))
(mf-sethash x
(cond ((<= (length str) n) str)
(t (concatenate
'string
(subseq str 0 (max 0 n))
"...")))
*mf-shorten-ht*)))))))))
(defun-one-output mf-print-alist (alist separation)
; Consider invoking this function in the context of our-syntax.
(check-type separation (integer 0))
(setq alist
(loop for x in alist collect
(progn
(check-type x
(cons (or string symbol)
(cons (or string (integer 0))
null)))
(list (mf-shorten (car x) *mf-print-alist-width*)
(if (integerp (cadr x))
(mf-num-to-string (cadr x))
(cadr x))))))
(let* ((max1 (loop for x in alist maximize (length (car x))))
(max2 (loop for x in alist maximize (length (cadr x))))
(width (max (or *print-right-margin*
*mf-print-right-margin*)
(+ separation max1 max2))))
(loop for x in alist do
(fresh-line)
(princ (car x))
(loop for i
below (- width (+ (length (car x))
(length (cadr x))))
do (write-char #\Space))
(princ (cadr x))))
nil)
(defmacro safe-incf-pons (x inc)
; This function was originally called very-unsafe-incf, and as its name
; implied, it was defined to increment x by inc without checking that the
; result is of mfixnum type. We are guessing that the cost of being safe is
; small, so we are renaming this function as safe-incf-pons, since it has only
; been used for reporting statistics related to pons. It would certainly be
; fine to eliminate this extra layer altogether once we are confident that it
; causes only negligible slowdown in statistics reporting.
`(safe-incf ,x ,inc))
(defun-one-output pons-summary ()
; Note: This function should probably either be made available in the loop or
; else eliminated.
; This function is not called anywhere else in the ACL2 sources, nor is it
; available from the loop. Perhaps though it could be useful in tuning the
; implementation.
(our-syntax
(let ((sssub 0) ; sum of sizes of subtables
(nponses 0) ; number of ponses
(nsubs 0) ; number of subsidiary hash tables (in flex alists)
(nponstab 0)) ; number of pons tables
(declare (type mfixnum sssub nponses nsubs))
(format t "Pons-summary:~%")
(with-global-memoize-lock
(mf-maphash
(lambda (k v)
(when (symbolp k)
(let ((tab (symbol-value (access memoize-info-ht-entry v
:ponstablename))))
(when tab
(safe-incf-pons nponstab 1)
(mf-maphash
(lambda (k v)
(declare (ignore k))
(cond
((not (listp v)) ; the flex-alist v is a hash table
(safe-incf-pons sssub (mf-hash-table-size v))
(safe-incf-pons nponses (mf-hash-table-count v))
(safe-incf-pons nsubs 1))
(t (safe-incf-pons nponses (length v)))))
tab)))))
*memoize-info-ht*))
(mf-print-alist
`((" Pons hits/calls"
,(let* ((c *pons-call-counter*)
(m *pons-misses-counter*)
(h (- c m)))
(format nil "~,1e / ~,1e = ~,2f" h c (if (eql c 0) 0 (/ h c)))))
(" Number of pons tables" ,(mf-num-to-string nponstab))
(" Number of pons sub tables" ,(mf-num-to-string nsubs))
(" Sum of pons sub table sizes" ,(mf-num-to-string sssub))
(" Number of ponses" ,(mf-num-to-string nponses)))
5)
(format t ")")
(force-output *standard-output*)
nil)))
(defun memoized-values (&optional (fn (memoized-functions)))
; Note: This function should probably either be made available in the loop or
; else eliminated.
; We print all the currently memoized values for the given function symbol or
; list of function symbols, which by default is the list of all memoized
; function symbols.
(with-global-memoize-lock
(cond ((listp fn) (mapc #'memoized-values fn))
((not (memoizedp-raw fn))
(format t "~%; Memoized-values: ~s is not memoized." fn))
(t (let ((tb (symbol-value
(access memoize-info-ht-entry
(mf-gethash fn *memoize-info-ht*)
:tablename))))
(cond ((and tb (not (eql 0 (mf-hash-table-count tb))))
(format t "~%; Memoized values for ~s." fn)
(mf-maphash (lambda (key v)
(format t "~%~s~%=>~%~s" key v))
tb)))))))
(force-output *standard-output*)
nil)
(defun print-call-stack ()
; Note: This function should probably either be made available in the loop or
; else eliminated.
; (Print-call-stack) prints the stack of memoized function calls currently
; running and the time they have been running.
(float-ticks/second-init)
(our-syntax
(let (lst
(ticks (internal-real-ticks))
(*print-case* :downcase))
(with-global-memoize-lock
(mf-maphash
(lambda (k v)
(when (symbolp k)
(let ((x (symbol-value (the symbol
(access memoize-info-ht-entry
v :start-ticks)))))
(declare (type mfixnum x))
(when (> x 0) ; in particular, on the stack (x not -1)
(push (cons k x) lst)))))
*memoize-info-ht*))
(setq lst (sort lst #'< :key #'cdr))
(setq lst (loop for pair in lst collect
(list (car pair)
(mf-num-to-string (/ (the-mfixnum
(- ticks (cdr pair)))
*float-ticks/second*)))))
(when lst
(terpri *standard-output*)
(mf-print-alist
(cons '("Stack of monitored function calls."
"Time since outermost call.")
lst)
5))))
(force-output *standard-output*)
nil)
(defun-one-output pons-calls (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (pons-calls x) is the number of times that x has
; called pons.
(setq x (coerce-index x))
(aref *memoize-call-array*
(ma-index x *ma-pons-index*)))
#+(or ccl sbcl)
(defun-one-output bytes-allocated (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (bytes-allocated x) is the number of
; heap bytes x has caused to be allocated on the heap.
(setq x (coerce-index x))
(aref *memoize-call-array*
(ma-index x *ma-bytes-index*)))
(defun-one-output number-of-hits (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (number-of-hits x) is the number of
; times that a call of x returned a remembered answer.
(setq x (coerce-index x))
(aref *memoize-call-array*
(ma-index x *ma-hits-index*)))
(defun-one-output number-of-memoized-entries (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (number-of-memoized-entries x) is the number of
; entries currently stored for x."
(setq x (coerce-index x))
(with-global-memoize-lock
(let ((h (mf-gethash x *memoize-info-ht*)))
(unless h (error "~a is not memoized." x))
(let* ((sym (access memoize-info-ht-entry
h
:tablename))
(val (symbol-value sym)))
(if (hash-table-p val) ; note: val is nil after clear-memoize-table
(mf-hash-table-count val)
0)))))
(defun-one-output number-of-mht-calls (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (number-of-mht-calls x) is the number
; of times that the memoize hash-table for x was created.
(setq x (coerce-index x))
(aref *memoize-call-array*
(ma-index x *ma-mht-index*)))
(defun-one-output total-time (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (total-time x) is the total time spent inside
; (all outermost) calls of x.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(float-ticks/second-init)
(/ (aref *memoize-call-array* (ma-index-ticks x))
*float-ticks/second*))
(defun-one-output number-of-calls (x)
; This function symbol can be included in *memoize-summary-order-list*.
; For a memoized function x, (number-of-calls x) is the total number of calls
; of x.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(aref *memoize-call-array*
(ma-index-calls x)))
(defun-one-output time-for-non-hits/call (x)
; This function symbol can be included in *memoize-summary-order-list*.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(let ((n (- (number-of-calls x) (number-of-hits x))))
(if (zerop n) 0 (/ (total-time x) n))))
(defun-one-output time/call (x)
; This function symbol can be included in *memoize-summary-order-list*.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(let ((n (number-of-calls x)))
(if (zerop n) 0 (/ (total-time x) n))))
(defun-one-output hits/calls (x)
; This function symbol can be included in *memoize-summary-order-list*.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(let ((n (number-of-calls x)))
(if (zerop n)
0
(/ (number-of-hits x)
; Why (float n) instead of n? We see no clear reason, but we leave this
; historical call of float in place.
(float n)))))
#+(or ccl sbcl)
(defun-one-output bytes-allocated/call (x)
; This function symbol can be included in *memoize-summary-order-list*.
; One must first call compute-calls-and-times to get sensible results.
(setq x (coerce-index x))
(let ((n (number-of-calls x)))
(if (zerop n) 0 (/ (bytes-allocated x) n))))
(defun char-list-fraction (l)
(if (atom l)
0
(+ (char-code (car l))
(/ (char-list-fraction (cdr l))
256))))
(defun symbol-name-order (s)
; Symbol-name-order maps symbols to rationals preserving lexicographic order.
; This function symbol can be included in *memoize-summary-order-list*.
(unless (symbolp s) (setq s (fixnum-to-symbol s)))
(- (char-list-fraction (coerce (symbol-name s) 'list))))
(defun-one-output execution-order (s)
; This function symbol can be included in *memoize-summary-order-list*.
(unless (symbolp s) (setq s (fixnum-to-symbol s)))
(the-mfixnum (symbol-value
(the symbol
(access memoize-info-ht-entry
(mf-gethash s *memoize-info-ht*)
:start-ticks)))))
(defun compute-calls-and-times ()
(with-global-memoize-lock ; perhaps not necessary for this, but harmless
(let ((ma *memoize-call-array*)
(2m *2max-memoize-fns*)
(ca *callers-array*)
(n ; fixnum, because *max-symbol-to-fixnum* <= 1/2 *2max-memoize-fns*
(the fixnum (1+ *max-symbol-to-fixnum*))))
(declare (type (simple-array mfixnum (*)) ma)
(type (simple-array t (*)) ca)
(type fixnum 2m n))
(cond ((eql (length ca) n)
(loop for i fixnum below n
do (setf (aref ca i) nil)))
(t (setq *callers-array* (make-array n :initial-element nil))
(setq ca *callers-array*)))
(loop for i fixnum below (ma-index-calls n 0 2m)
; fixnum, because (ma-index-calls n 0 2m) = 2n <= *2max-memoize-fns*
; Initialize column 0 of *memoize-call-array* to set to 0, for each i < n, both
; (ma-index-calls i 0 2m) and (ma-index-ticks i 0 2m).
do (setf (aref ma i) 0))
(loop for i fixnum ; column, for caller
from *initial-max-symbol-to-fixnum*
to *max-symbol-to-fixnum* ; indeed a fixnum; see comments above
do
(let ((col-base (ma-index i 0 2m)))
(declare (type mfixnum col-base))
(loop for j fixnum ; row
from *initial-max-symbol-to-fixnum*
to *max-symbol-to-fixnum* ; indeed a fixnum; see above
do
(let* ((calls-row (ma-index-calls j))
(index (ma-index-from-col-base col-base calls-row))
(calls (the-mfixnum (aref ma index))))
(declare (type mfixnum calls-row index calls))
(when (> calls 0)
(let* ((ticks-index
; The index for ticks is (ma-index-from-col-base col-base (ma-index-ticks j)).
; But that is just (1+ index), so we use the latter, more efficient form.
(the-mfixnum (1+ index)))
(ticks (aref ma ticks-index))
(ticks-row ; (ma-index-ticks j)
(the-mfixnum (1+ calls-row))))
(declare (type mfixnum ticks-index ticks ticks-row))
(safe-incf (aref ma calls-row)
calls
compute-calls-and-times)
(safe-incf (aref ma ticks-row)
ticks
compute-calls-and-times)
(push i ; Add i to the callers of j.
(aref ca j)))))))))))
(defun-one-output print-not-called ()
; Note: This function should probably either be made available in the loop or
; else eliminated.
; (Print-not-called) prints, one to a line, in alphabetic order, the currently
; memoized functions that have never been called. This utility may be useful
; when looking for test coverage.
(with-global-memoize-lock
(compute-calls-and-times)
(let ((ans nil))
(mf-maphash (lambda (k v)
(declare (ignore v))
(when (and (symbolp k)
(eql 0 (number-of-calls k)))
(push k ans)))
*memoize-info-ht*)
(loop for x in (sort ans 'alphorder)
do (print x)))))
(defun-one-output memoize-summary-sort ()
(labels
((lex-> (l1 l2)
(cond ((or (atom l1)
(atom l2))
nil)
((> (car l1) (car l2)) t)
((< (car l1) (car l2)) nil)
(t (lex-> (cdr l1) (cdr l2))))))
(let (pairs)
(with-global-memoize-lock
(mf-maphash
(lambda (fn v)
(when (symbolp fn)
(let ((num (access memoize-info-ht-entry v :num)))
(declare (type mfixnum num))
(when (< 0 (number-of-calls num))
(push (cons fn (loop for order
in *memoize-summary-order-list*
collect (funcall order fn)))
pairs)))))
*memoize-info-ht*)
(sort pairs
(if *memoize-summary-order-reversed*
(lambda (x y) (lex-> y x))
#'lex->)
:key #'cdr)))))
(defun-one-output outside-caller-p (x)
(with-global-memoize-lock
(or (<= x *initial-max-symbol-to-fixnum*)
(null (mf-gethash x *memoize-info-ht*)))))
(defun-one-output memoize-condition (fn)
(with-global-memoize-lock
(let ((x (mf-gethash fn *memoize-info-ht*)))
(and x
(access memoize-info-ht-entry x :condition)))))
(defun-one-output memoize-summary-after-compute-calls-and-times ()
; One must first call compute-calls-and-times to get sensible results.
(with-global-memoize-lock ; perhaps not necessary for this, but harmless
(float-ticks/second-init)
(our-syntax
(let* ((fn-pairs (memoize-summary-sort))
(ma *memoize-call-array*)
(len-orig-fn-pairs (len fn-pairs))
(len-fn-pairs 0) ; set below
#+(or ccl sbcl)
(global-bytes-allocated 0) ; set below
(global-pons-calls 0) ; set below
)
(declare (type (simple-array mfixnum (*)) ma)
(type fixnum len-orig-fn-pairs len-fn-pairs))
(when (and *memoize-summary-limit*
(> len-orig-fn-pairs *memoize-summary-limit*))
(setq fn-pairs (take *memoize-summary-limit* fn-pairs)))
(setq len-fn-pairs (len fn-pairs))
(when (> len-fn-pairs 0)
(format t
"~%Sorted by *memoize-summary-order-list* = ~s."
*memoize-summary-order-list*)
(when (< len-fn-pairs len-orig-fn-pairs)
(format t
"~%Reporting on ~:d of ~:d functions because ~
*memoize-summary-limit* = ~a."
len-fn-pairs
len-orig-fn-pairs
*memoize-summary-limit*)))
#+(or ccl sbcl)
(setq global-bytes-allocated
(loop for pair in fn-pairs sum
(bytes-allocated (car pair))))
(setq global-pons-calls
(loop for pair in fn-pairs sum (pons-calls (car pair))))
(when (null fn-pairs)
(format t "~%(memoize-summary) has nothing to report.~%"))
(loop for pair in fn-pairs do
(let* ((fn (car pair))
(entry (mf-gethash fn *memoize-info-ht*))
(tablename
(symbol-value
(access memoize-info-ht-entry entry :tablename)))
(ponstablename
(symbol-value
(access memoize-info-ht-entry entry :ponstablename)))
(start-ticks
(the-mfixnum
(symbol-value
(the symbol
(access memoize-info-ht-entry entry
:start-ticks)))))
(num (the-mfixnum (access memoize-info-ht-entry entry
:num)))
(nhits (the-mfixnum (number-of-hits num)))
(nmht (the-mfixnum (number-of-mht-calls num)))
(ncalls (the-mfixnum (number-of-calls num)))
(pons-calls (the-mfixnum (pons-calls num)))
(no-hits (or (not *report-hits*)
(null (memoize-condition fn))))
#+(or ccl sbcl)
(bytes-allocated (bytes-allocated num))
(tt (max .000001
; Why insist that tt be at least .000001? That has been the case historically,
; and it might be reasonable since a smaller time might be unlikely and anyhow,
; .000001 is not unreasonable if only as "noise". So we leave this code alone.
(total-time num)))
(t/c (time/call num))
(in-progress-str
(if (eql start-ticks -1) "" " (running)"))
(selftime tt))
(declare (type integer start-ticks)
(type mfixnum num nhits nmht ncalls
pons-calls
#+(or ccl sbcl) bytes-allocated))
(format t "~%(~s~%" fn)
(mf-print-alist
`(,@(when (or *report-calls* *report-hits*)
`((,(format nil " ~a~a"
(if no-hits "Calls" "Hits/calls")
in-progress-str)
,(cond
(no-hits (format nil "~a" (mf-num-to-string ncalls)))
((zerop ncalls)
(format nil "~8,2e/~8,2e = ????%"
nhits
0))
(t (format nil "~8,2e/~8,2e = ~4,1f%"
nhits
ncalls
(* 100 (/ nhits (float ncalls)))))))))
,@(when (and *report-mht-calls*
; The following inequality was here historically, perhaps because nmht is
; always at least 1 (not sure about that); anyhow, it seems OK to leave it.
(>= nmht 2))
`((" Number of calls to mht" ,(mf-num-to-string nmht))))
,@(when *report-time*
`((" Time of all outermost calls"
,(format nil "~a"
(mf-num-to-string tt)))
(" Time per call" ,(mf-num-to-string t/c))))
; The commented-out code (when ...), below, has been here historically (through
; August 2014), but it's not clear why. Note that the "Heisenberg" note is
; only made for profiled functions; what's so special about the profiling case?
; Let's just leave it to the user to decide what to make of a very small time
; per call. To see such a case, you can try this:
; (defn foo (x) x)
; (profile 'foo)
; (foo 1)
; (foo 2)
; (memsum)
; ,@(when (and (null (access memoize-info-ht-entry entry
; :condition))
; (null (access memoize-info-ht-entry entry
; :inline))
; (< t/c 1e-6))
; `((,(format nil " Doubtful timing info for ~a." fn)
; "Heisenberg effect.")))
#+(or ccl sbcl)
,@(when (and (> bytes-allocated 0) *report-bytes*)
(assert (> global-bytes-allocated 0))
`((" Heap bytes allocated"
,(format nil "~a; ~4,1f%"
(mf-num-to-string bytes-allocated)
(* 100 (/ bytes-allocated
global-bytes-allocated))))
(" Heap bytes allocated per call"
,(mf-num-to-string (/ bytes-allocated ncalls)))))
,@(when (and (> pons-calls 0)
*report-pons-calls*)
(assert (> global-pons-calls 0))
`((" Pons calls"
,(format nil "~a; ~4,1f%"
(mf-num-to-string pons-calls)
(* 100 (/ pons-calls global-pons-calls))))))
,@(when (and *report-hits* *report-time* (not (eql 0 nhits)))
`((" Time per missed call"
,(mf-num-to-string (time-for-non-hits/call num)))))
,@(when *report-calls-to*
(let ((lst nil)
(outside-fn-time 0)
(outside-fn-count 0))
(declare (type mfixnum outside-fn-count))
(loop
for callern in ; collect fns called by fn
(loop for i below (length *callers-array*)
when (member num (aref *callers-array* i))
collect i)
do
(let* ((call-loc (ma-index-calls callern num))
(calls (aref ma call-loc))
(ticks-loc ; (ma-index-ticks callern num)
(the-mfixnum (1+ call-loc)))
(ticks (aref ma ticks-loc))
(local-tt (/ ticks *float-ticks/second*)))
(declare
(type mfixnum call-loc calls ticks-loc ticks))
(decf selftime local-tt)
(cond
((or (outside-caller-p callern)
(< (* 100 local-tt) tt))
(incf outside-fn-time local-tt)
(safe-incf
outside-fn-count
calls
memoize-summary-after-compute-calls-and-times))
(t (push
`((,(format nil
" To ~s"
(fixnum-to-symbol callern))
,(format nil
"~8,2e calls took~8,2e; ~a%"
calls
local-tt
(cond
((> (* 10000 local-tt) tt)
(assert (< 0 tt))
(format
nil
"~4,1f"
(* 100 (/ local-tt tt))))
(t "<0.01"))))
. ,(if *sort-to-from-by-calls*
calls
local-tt))
lst)))))
(when (> outside-fn-time 0)
(assert (< 0 tt))
(push
`((,(format nil " To other functions")
,(format
nil
"~8,2e calls took~8,2e; ~a%"
outside-fn-count
outside-fn-time
(if (> (* 10000 outside-fn-time) tt)
(format nil
"~4,1f"
(* 100 (/ outside-fn-time tt)))
"<0.01")))
. ,(if *sort-to-from-by-calls*
outside-fn-count
outside-fn-time))
lst))
(when (and (> selftime 0)
(not (= selftime tt)))
(assert (< 0 tt))
(push `((" To self/unprofiled functions"
,(format
nil
"~8,2e; ~a%"
selftime
(if (> (* 10000 selftime) tt)
(format nil
"~4,1f"
(* 100 (/ selftime tt)))
"<0.01")))
. ,(if *sort-to-from-by-calls*
0 ; unknown number of calls; just use 0
selftime))
lst))
(setq lst (sort lst #'> :key #'cdr))
(strip-cars lst)))
,@(when *report-calls-from*
(let ((lst nil)
(outside-caller-time 0)
(outside-caller-count 0))
(loop
for callern in (aref *callers-array* num)
do
(let* ((call-loc (ma-index-calls num callern))
(calls (aref ma call-loc))
(ticks-loc ; (ma-index-ticks num callern)
(the-mfixnum (1+ call-loc)))
(ticks (aref ma ticks-loc))
(local-tt (/ ticks *float-ticks/second*)))
(declare
(type mfixnum call-loc calls ticks-loc ticks))
(cond
((or (outside-caller-p callern)
(< (* 100 local-tt) tt))
(incf outside-caller-time local-tt)
(incf outside-caller-count calls))
(t (push
`((,(format
nil
" From ~s"
(fixnum-to-symbol callern))
,(format
nil
"~8,2e calls took~8,2e; ~a%"
calls
local-tt
(if (and (<= .001 local-tt)
(<= local-tt tt))
(format nil
"~4,1f"
(* 100 (/ local-tt tt)))
"?")))
. ,(if *sort-to-from-by-calls*
calls
local-tt))
lst)))))
(when (> outside-caller-time 0)
(push
`((,(format nil " From other functions")
,(format
nil
"~8,2e calls took~8,2e; ~a%"
outside-caller-count
outside-caller-time
(if (and (<= .001 outside-caller-time)
(<= outside-caller-time tt))
(format
nil
"~4,1f"
(* 100 (/ outside-caller-time tt)))
"?")))
. ,(if *sort-to-from-by-calls*
outside-caller-count
outside-caller-time))
lst))
(setq lst (sort lst #'> :key #'cdr))
(strip-cars lst)))
,@(when (or *report-on-memo-tables* *report-on-pons-tables*)
(let ((sum-pons-sub-sizes 0)
(nponses 0)
(nponssubs 0))
(declare (type mfixnum
sum-pons-sub-sizes nponses nponssubs))
(and
ponstablename
*report-on-pons-tables*
(mf-maphash
(lambda (key value)
(declare (ignore key))
(cond
((not (listp value))
(safe-incf-pons sum-pons-sub-sizes
(mf-hash-table-size value))
(safe-incf-pons nponses
(mf-hash-table-count value))
(safe-incf-pons nponssubs 1))
(t (safe-incf-pons nponses (length value)))))
ponstablename))
`(,@(and
tablename
*report-on-memo-tables*
(let ((count (mf-hash-table-count tablename))
(size (mf-hash-table-size tablename)))
`((,(format nil " Memoize table count/size")
,(format
nil
"~8,2e/~8,2e = ~4,1f%"
count size (* 100 (/ count size)))))))
,@(and
ponstablename
*report-on-pons-tables*
(let ((count (mf-hash-table-count ponstablename))
(size (mf-hash-table-size ponstablename)))
`((" (Pons table count/size"
,(format
nil
"~:d/~:d = ~4,1f%)"
count
size
(* 100 (/ count size))))
(" (Number of pons sub tables"
,(format
nil
"~a)"
(mf-num-to-string nponssubs)))
(" (Sum of pons sub table sizes"
,(format
nil
"~a)"
(mf-num-to-string sum-pons-sub-sizes)))
(" (Number of ponses"
,(format
nil
"~a)"
(mf-num-to-string nponses))))))))))
5))
(format t ")"))))
(terpri *standard-output*)
(force-output *standard-output*))
nil)
(defun-one-output memoize-summary ()
; Warning: Keep the return values in sync for the logic and raw Lisp.
; (Memoize-summary) reports data stored during the execution of the functions
; in (memoized-functions).
; Typically each call of a memoized function, fn, is counted. The elapsed time
; until an outermost function call of fn ends, the number of heap bytes
; allocated in that period (CCL and SBCL only), and other 'charges' are
; 'billed' to fn. That is, quantities such as elapsed time and heap bytes
; allocated are not charged to subsidiary recursive calls of fn while an
; outermost call of fn is running. Recursive calls of fn, and memoized 'hits',
; are counted, unless fn was memoized with nil as the value of the :inline
; parameter of memoize.
; The settings of the following determine, at the time a function is
; given to memoize, the information that is collected for calls of
; the function:
; Variable type
; *record-bytes* boolean (available in CCL and SBCL only)
; *record-calls* boolean
; *record-hits* boolean
; *record-mht-calls* boolean
; *record-pons-calls* boolean
; *record-time* boolean
; The settings of the following determine, at the time that
; memoize-summary is called, what information is printed:
; *report-bytes* boolean (available in CCL and SBCL only)
; *report-calls* boolean
; *report-calls-from* boolean
; *report-calls-to* boolean
; *report-hits* boolean
; *report-mht-calls* boolean
; *report-pons-calls* boolean
; *report-time* boolean
; *report-on-memo-tables* boolean
; *report-on-pons-tables* boolean
; *memoize-summary-limit* (or integerp null)
; *memoize-summary-order-list* (symbol symbol ... symbol)
; *memoize-summary-order-reversed* boolean
; Functions are sorted lexicographically according to the ordering
; induced by the function names in *memoize-summary-order-list*, each
; member of which must be a unary function that returns a rational.
; (Clear-memoize-tables) forgets the remembered values of all memoized function
; calls. It does not alter a function's status as being a memoized function,
; nor does it alter the monitoring data accumulated.
; (Unmemoize-all) undoes the memoization status of all memoized
; functions.
; (Clear-memoize-call-array) zeroes out the monitoring information for
; all functions. It does not alter any function's status as a
; memoized function nor does it change the values remembered for any
; memoized function.
; Here is an example of hacking with *memoize-summary-order-list* that
; instructs memoize-summary to print information about, say,
; 1st-mod-err first:
; (push (lambda (fn)
; (if (eq fn '1st-mod-err) 1 0))
; *memoize-summary-order-list*)."
; Note that like the "Time of all outermost calls", the "Pons calls" reported
; are cumulative. For example, we get 8 pons calls for G (not 4) in the
; following example, using CCL: 4 for G, and 4 for the subsidiary calls of F.
; (defn f (x y) (cons x y))
; (defn g (x y) (f x y))
; (memoize 'f)
; (memoize 'g)
; (g '(3 . 4) '(5 . 6))
; (g '(3 . 5) '(5 . 6))
; (g '(3 . 6) '(5 . 6))
; (g '(3 . 7) '(5 . 6))
; (memsum)
(compute-calls-and-times)
(memoize-summary-after-compute-calls-and-times)
nil)
; SECTION: Clearing and initialization
(defun clear-one-memo-and-pons-hash (l)
; It is debatable whether one should use the clrhash approach or the
; set-to-nil approach in clear-one-memo-and-pons-hash. The clrhash approach,
; in addition to reducing the number of make-hash-table calls necessary, has
; the effect of immediately clearing a hash-table even if some other function
; is holding on to it, so more garbage may get garbage collected sooner than
; otherwise. The set-to-nil approach has the advantage of costing very few
; instructions and very little paging. See also the comment "Perhaps we
; should use clrhash below" in memoize-fn-outer-body.
(with-global-memoize-lock
(let* ((fn (access memoize-info-ht-entry l :fn))
(mt (symbol-value (access memoize-info-ht-entry l :tablename)))
(pt (symbol-value (access memoize-info-ht-entry l :ponstablename)))
(mt-count (and mt (mf-hash-table-count mt)))
(pt-count (and pt (mf-hash-table-count pt))))
(when mt
(setf (symbol-value (access memoize-info-ht-entry l :tablename))
nil))
(when pt
(setf (symbol-value (access memoize-info-ht-entry l :ponstablename))
nil))
(when (or mt-count pt-count)
(update-memo-max-sizes fn (or pt-count 1) (or mt-count 1)))))
nil)
(defun-one-output clear-memoize-table (k)
; Warning: Keep the return values in sync for the logic and raw Lisp.
(with-global-memoize-lock
(when (symbolp k)
(let ((l (mf-gethash k *memoize-info-ht*)))
(when l (clear-one-memo-and-pons-hash l)))))
k)
(defun-one-output clear-memoize-tables ()
; Warning: Keep the return values in sync for the logic and raw Lisp.
(with-global-memoize-lock
(let (success)
(unwind-protect
(progn
(mf-maphash (lambda (k l)
(when (symbolp k)
(clear-one-memo-and-pons-hash l)))
*memoize-info-ht*)
(setq success t))
(or success (error "clear-memoize-tables failed."))))
(setq *pons-call-counter* 0)
(setq *pons-misses-counter* 0))
nil)
(defun clear-memoize-call-array ()
; (Clear-memoize-call-array) zeros out all records of function calls for
; purposes of reporting, for example by memoize-summary.
(with-global-memoize-lock
(let ((ma *memoize-call-array*))
(declare (type (simple-array mfixnum (*)) ma))
(let ((len (length ma)))
(cond ((typep len 'fixnum)
(loop for i fixnum below len
do (setf (aref ma i) 0)))
(t ; maybe impossible?
(loop for i of-type integer below len
do (setf (aref ma i) 0))))))))
(defun clear-memoize-statistics ()
; Warning: Keep the return values in sync for the logic and raw Lisp.
(with-global-memoize-lock
(setq *pons-call-counter* 0)
(setq *pons-misses-counter* 0)
(clear-memoize-call-array))
nil)
(defun-one-output memoize-init ()
; This function initializes most globals that are managed automatically by the
; memoization implementation, i.e., that are not user-settable. Not included
; here is *sol-gc-installed*, which we consider to pertain to ACL2(h) but to be
; outside memoization proper.
(with-global-memoize-lock
(when *memoize-init-done*
(return-from memoize-init nil))
(unless (eql *caller* (the-mfixnum (outside-caller-col-base)))
(error "memoize-init: A memoized function is running."))
(let (success)
(unwind-protect
(progn
(setq *pons-call-counter* 0)
(setq *pons-misses-counter* 0)
(setq *memoize-info-ht* (initial-memoize-info-ht))
(mf-sethash *initial-max-symbol-to-fixnum*
"outside-caller"
*memoize-info-ht*)
(setq *max-symbol-to-fixnum* *initial-max-symbol-to-fixnum*)
(setq *2max-memoize-fns* *initial-2max-memoize-fns*)
(sync-memoize-call-array)
(setq *memo-max-sizes* (mf-mht :shared :default))
(setq success t))
(if success
(setq *memoize-init-done* t)
(error "memoize-init failed.")))))
nil)
(defmacro with-lower-overhead (&rest r)
; Warning: Keep this in sync with lower-overhead and with-higher-overhead.
`(let ((*record-bytes* nil)
(*record-calls*
; It has been a mystery why we have needed to set *record-calls* to t in
; LispWorks. We found that otherwise, for example, evaluation hangs for
; (bad-lisp-objectp (make-list 100000)) when bad-lisp-objectp is in the initial
; memoized state produced by calling acl2h-init (see acl2h-init-memoizations)
; It might well be worth looking into this again.
#-lispworks nil #+lispworks t)
(*record-hits* nil)
(*record-mht-calls* nil)
(*record-pons-calls* nil)
(*record-time* nil))
,@r))
(defmacro with-higher-overhead (&rest r)
; Warning: Keep this in sync with lower-overhead and with-lower-overhead.
`(let ((*record-bytes* t)
(*record-calls* t)
(*record-hits* t)
(*record-mht-calls* t)
(*record-pons-calls* t)
(*record-time* t))
,@r))
(defmacro with-overhead (val form)
(let ((v (gensym)))
`(let ((,v ,val))
(cond ((eq ,v :default) ,form)
((null ,v) (with-lower-overhead ,form))
(t (with-higher-overhead ,form))))))
(defun acl2h-init-memoizations ()
; The memoizations performed here, for certain built-in functions, may be
; important for performance in applications that traffic in large objects or
; terms. Some of these memoizations are undone by acl2h-init-unmemoizations,
; which is called by ACL2(hp) when waterfall-parallelism is turned on, in order
; to avoid errors during the waterfall due the fact that memoization is not
; thread-safe.
(loop for entry in
*thread-unsafe-builtin-memoizations*
when (not (memoizedp-raw (car entry)))
do (with-lower-overhead
(apply 'memoize-fn entry))))
;;;;;;;;;;
;;; Start memory management code (start-sol-gc)
;;;;;;;;;;
; This section of code was suggested by Jared Davis as a way to regain
; performance of ACL2(h) on regressions at UT CS. Initially, these regressions
; showed significant slowdown upon including new memoization code from Centaur
; on 3/28/2013:
; ; old:
; 24338.570u 1357.200s 1:19:02.75 541.7% 0+0k 0+1918864io 0pf+0w
; ; new:
; 33931.460u 1017.070s 1:43:24.28 563.2% 0+0k 392+1931656io 0pf+0w
; After restoring (start-sol-gc) in function acl2h-init, we regained the old
; level of performance for a UT CS ACL2(h) regression, with the new memoizaion
; code.
(defun mf-looking-at (str1 str2 &key (start1 0) (start2 0))
; (Mf-looking-at str1 str2 :start1 s1 :start2 s2) is non-nil if and only if
; string str1, from location s1 to its end, is an initial segment of string
; str2, from location s2 to its end.
(unless (typep str1 'simple-base-string)
(error "looking at: ~a is not a string." str1))
(unless (typep str2 'simple-base-string)
(error "looking at: ~a is not a string." str2))
(unless (typep start1 'fixnum)
(error "looking at: ~a is not a fixnum." start1))
(unless (typep start2 'fixnum)
(error "looking at: ~a is not a fixnum." start2))
(locally
(declare (simple-base-string str1 str2)
(fixnum start1 start2))
(let ((l1 (length str1)) (l2 (length str2)))
(declare (fixnum l1 l2))
(loop
(when (>= start1 l1) (return t))
(when (or (>= start2 l2)
(not (eql (char str1 start1)
(char str2 start2))))
(return nil))
(incf start1)
(incf start2)))))
(defun our-uname ()
; Returns nil or else a keyword, currently :darwin or :linux, to indicate the
; result of shell command "uname".
(multiple-value-bind
(exit-code val)
(system-call+ "uname" nil)
(and (eql exit-code 0)
(stringp val)
(<= 6 (length val))
(cond ((string-equal (subseq val 0 6) "Darwin") :darwin)
((string-equal (subseq val 0 5) "Linux") :linux)))))
(defun meminfo (&optional arg)
; With arg = nil, this function either returns 0 or else the size of the
; physical memory. See the code below to understand what information might be
; returned for non-nil values of arg.
(assert (or (null arg)
(stringp arg)))
(or
(with-standard-io-syntax
(case (our-uname)
(:linux
(let ((arg (or arg "MemTotal:")))
(and
(our-ignore-errors (probe-file "/proc/meminfo"))
(with-open-file
(stream "/proc/meminfo")
(let (line)
(loop while (setq line (read-line stream nil nil)) do
(when (mf-looking-at arg line)
(return
(values
(read-from-string line nil nil
:start (length arg)))))))))))
(:darwin
(let* ((arg (or arg "hw.memsize"))
(len (length arg)))
(multiple-value-bind
(exit-code val)
(system-call+ "sysctl" (list arg))
(and (eql exit-code 0)
(mf-looking-at arg val)
(mf-looking-at arg ": " :start1 len)
(let ((ans (read-from-string val nil nil :start (+ 2 len))))
(and (integerp ans)
(equal (mod ans 1024) 0)
(/ ans 1024)))))))
(t nil)))
0))
(let ((physical-memory-cached-answer nil))
(defun physical-memory () ; in KB
(or physical-memory-cached-answer
(setq physical-memory-cached-answer
(meminfo))))
)
#+ccl
(defun set-and-reset-gc-thresholds ()
; See start-sol-gc for a full discussion. The comments here summarize how that
; works out if, for example, there are 8G bytes of physical memory, just to
; make the concepts concrete -- so it might be helpful to read the comments in
; this function before reading the more general discussion in start-sol-gc.
(let ((n
; E.g., with 8G bytes of physical memory, *max-mem-usage* is 1/8 of that --
; i.e., 1G -- and *gc-min-threshold* is 1/4 of that -- i.e., (1/4)G. Then here
; we arrange to allocate enough memory after a GC to reach *max-mem-usage* = 1G
; bytes before the next GC, unless the current memory usage is more than
; (3/4)G, in which case we allocate the minimum of (1/4)G.
(max (- *max-mem-usage* (ccl::%usedbytes))
*gc-min-threshold*)))
; Now set the "threshold" to the number of bytes computed above (unless that
; would be a no-op).
(unless (eql n (ccl::lisp-heap-gc-threshold))
(ccl::set-lisp-heap-gc-threshold n)))
; The above setting won't take effect until the next GC unless we take action.
; Here is that action, which actually allocates the bytes computed above as
; free memory.
(ccl::use-lisp-heap-gc-threshold)
; Finally, still assuming 8G bytes of phyical memory, set the "threshold" to
; (1/4)G. This is how much the next GC will set aside as free memory -- at
; least initially, but then the post-gc hook will call this function. As
; explained above, in the case that the current memory usage is less than
; (3/4)G, enough free memory will be allocated so that the next GC is triggered
; after *max-mem-usage* = 1G bytes are in use.
(unless (eql *gc-min-threshold* (ccl::lisp-heap-gc-threshold))
(ccl::set-lisp-heap-gc-threshold *gc-min-threshold*)))
#+ccl
(defun start-sol-gc ()
; Sol Swords's scheme to control GC in CCL
;
; The goal is to get CCL to perform a GC whenever we're using almost
; all the physical memory, but not otherwise.
;
; The discussion below is self-contained, but for more discussion, relevant CCL
; documentation is at http://ccl.clozure.com/ccl-documentation.html, Chapter
; 16.
;
; The usual way of controlling GC on CCL is via LISP-HEAP-GC-THRESHOLD. This
; value is approximately the amount of free memory that will be allocated
; immediately after GC. This means that the next GC will occur after
; LISP-HEAP-GC-THRESHOLD more bytes are used (by consing or array allocation or
; whatever). But this means the total memory used by the time the next GC
; comes around is the threshold plus the amount that remained in use at the end
; of the previous GC. This is a problem because of the following scenario:
;
; - We set the LISP-HEAP-GC-THRESHOLD to 3GB since we'd like to be able
; to use most of the 4GB physical memory available.
;
; - A GC runs or we say USE-LISP-HEAP-GC-THRESHOLD to ensure that 3GB
; is available to us.
;
; - We run a computation until we've exhausted this 3GB, at which point
; a GC occurs.
;
; - The GC reclaims 1.2 GB out of the 3GB used, so there is 1.8 GB
; still in use.
;
; - After GC, 3GB more is automatically allocated -- but this means we
; won't GC again until we have 4.8 GB in use, meaning we've gone to
; swap.
;
; What we really want is, instead of allocating a constant additional
; amount after each GC, to allocate up to a fixed total amount including
; what's already in use. To emulate that behavior, we use the hack
; below. This operates as follows, assuming the same 4GB total physical
; memory as in the above example (or, unknown physical memory that defaults to
; 4GB as shown below, i.e., when function meminfo returns 0).
;
; 1. We set the LISP-HEAP-GC-THRESHOLD to (0.5G minus used bytes) and call
; USE-LISP-HEAP-GC-THRESHOLD so that our next GC will occur when we've used a
; total of 0.5G.
;
; 2. We set the threshold back to *gc-min-threshold*= 0.125GB without calling
; USE-LISP-HEAP-GC-THRESHOLD.
;
; 3. Run a computation until we use up the 0.5G and the GC is called. Say the
; GC reclaims 0.3GB so there's 0.2GB in use. 0.125GB more (the current
; LISP-HEAP-GC-THRESHOLD) is allocated so the ceiling is 0.325GB.
;
; 4. A post-GC hook runs which again sets the threshold to (0.5G minus used
; bytes), calls USE-LISP-HEAP-GC-THRESHOLD to raise the ceiling to 0.5G, then
; sets the threshold back to 0.125GB, and the process repeats.
;
; A subtlety about this scheme is that post-GC hooks runs in a separate
; thread from the main execution. A possible bug is that in step 4,
; between checking the amount of memory in use and calling
; USE-LISP-HEAP-GC-THRESHOLD, more memory might be used up by the main
; execution, which would set the ceiling higher than we intended. To
; prevent this, we interrupt the main thread to run step 4.
; The following settings are highly heuristic. We arrange that gc
; occurs at 1/8 of the physical memory size in bytes, in order to
; leave room for the gc point to grow (as per
; set-and-reset-gc-thresholds). If we can determine the physical
; memory; great; otherwise we assume that it it contains at least 4GB,
; a reasonable assumption we think for anyone using the HONS version
; of ACL2.
(let* ((phys (physical-memory))
(memsize (cond ((> phys 0) (* phys 1024)) ; to bytes
(t (expt 2 32)))))
(setq *max-mem-usage* (min (floor memsize 8)
(expt 2 31)))
(setq *gc-min-threshold* (floor *max-mem-usage* 4)))
(unless *sol-gc-installed*
(ccl::add-gc-hook
#'set-and-reset-gc-thresholds
:post-gc)
(setq *sol-gc-installed* t))
(set-and-reset-gc-thresholds))
(defun-one-output acl2h-init ()
; ACL2-DEFAULT-RESTART is called whenever a saved image for any version of ACL2
; is started up. For ACL2(h), ACL2-DEFAULT-RESTART calls ACL2H-INIT. Although
; we expect ACL2-DEFAULT-RESTART to be called only once, nevertheless for
; robustness we code ACL2H-INIT so that it may be called multiple times.
; (memoize-init) ; skipped, since this call is in exit-boot-strap-mode
#+static-hons
(setq *print-array*
; With *print-array* turned on, we end up sometimes seeing the SBITS array in
; backtraces, etc, which can effectively kill your session (can't interrupt,
; etc.). This is only a problem with static-honsing since classic honsing
; doesn't have sbits anyway.
nil)
#+(and ccl (not mswindows))
(start-sol-gc)
; We have decided not to print garbage collector messages, since these can be
; distracting to ordinary users. Those who want such messages can put
; (gc-verbose t), or even (gc-verbose t t), in their ~/acl2-customization.lsp
; files. We queried the acl2-help mailing list on [16 Jan 2015 17:34:36] and
; gave people the opportunity to object (albeit not particularly inviting such
; objections), but nobody did. We keep the following historical comment (and
; code), which provides insight for why this decision required thought. Note
; that gc messages are turned off, where we know how to do so, in source file
; acl2.lisp.
; It is can be helpful for the user to know what the garbage collector is
; doing when using HONS and MEMOIZE. We leave full-gc messages on but turn
; EGC messages off. We're disabling EGC anyway, but when we've experimented
; with leaving it on, the EGC messages are way too verbose.
; #+(or ccl cmu gcl) ; Lisps where gc-verbose is supported
; (gc-verbose t nil)
; We turn off EGC because it doesn't seem to work well with memoizing
; worse-than-builtin and sometimes seems buggy; but we want to investigate this
; more.
#+ccl
(ccl::egc nil)
nil)
; SECTION: Miscellaneous (e.g. shorter, older names)
; Note: memsum is defined in memoize.lisp.
(defun memstat (&rest r)
(apply #'memoized-values r))
(defun clear-memo-tables ()
(clear-memoize-tables))
(defun lower-overhead ()
; Warning: Keep this in sync with with-lower-overhead and
; with-higher-overhead.
; An old comment here claims that lower-overhead does not help much (in
; speeding things up, presumably). It might be interesting to test that.
(setq *record-bytes* nil)
(setq *record-calls*
; See the comment about LispWorks in with-lower-overhead; we make the analogous
; adjustment for LispWorks here, in case it's necessary.
#-lispworks nil #+lispworks t)
(setq *record-hits* nil)
(setq *record-mht-calls* nil)
(setq *record-pons-calls* nil)
(setq *record-time* nil))
(defun update-memo-entry-for-attachments (fns entry wrld)
; See the Essay on Memoization with Attachments.
; This function is called by update-memo-entries-for-attachments. We return
; (mv changed-p new-entry), where if changed-p is not t or nil then it is a
; function symbol whose attachment has changed, which requires clearing of the
; corresponding memo table.
(let* ((ext-anc-attachments
(access memoize-info-ht-entry entry :ext-anc-attachments))
(valid-p
(if (eq fns :clear)
:clear
(or (null ext-anc-attachments)
(ext-anc-attachments-valid-p fns ext-anc-attachments wrld
t)))))
(cond ((eq valid-p t) (mv nil entry))
(t
(mv (if (eq valid-p nil) t valid-p)
(change memoize-info-ht-entry entry
:ext-anc-attachments
(ext-ancestors-attachments
(access memoize-info-ht-entry entry :fn)
wrld)))))))
(defun update-memo-entries-for-attachments (fns wrld state)
; See the Essay on Memoization with Attachments. Here is a brief, high-level
; summary of what is going on.
; This function is called by update-wrld-structures, which is called when the
; world is updated. When a defattach event is added or removed from the world,
; variable *defattach-fns* collects the name of the function with an attachment
; (either being added or removed). Ultimately, *defattach-fns* is passed as
; the fns parameter to the present function, and functions' entries in
; *memoize-info-ht* are correspondingly updated as necessary, in particular
; clearing tables whose values may have depended on attachments that are no
; longer present.
(let ((ctx 'top-level)
(fns (if (eq fns :clear)
:clear
(strict-merge-sort-symbol-<
(loop for fn in fns
collect (canonical-sibling fn wrld))))))
(when (eq fns :clear)
(observation ctx
"Memoization tables for functions memoized with :AOKP T ~
are being cleared."))
(when fns ; optimization
(with-global-memoize-lock
(mf-maphash
(lambda (k entry)
(when (symbolp k)
(mv-let (changedp new-entry)
(update-memo-entry-for-attachments fns entry wrld)
(when changedp
(when (not (or (eq changedp t)
(eq fns :clear)))
(observation ctx
"Memoization table for function ~x0 is ~
being cleared because attachment to ~
function ~x1 has changed."
k changedp)
(clear-one-memo-and-pons-hash entry))
(mf-sethash k new-entry *memoize-info-ht*)))))
*memoize-info-ht*)))))
|