Hi all
As my A.I. lecturer seems to think that advances in chess engine strength are largely due to improvements in hardware, I was wondering if anyone could point me to somewhere that evaluates the relative contribution of search and evaluation vs hardware to improvements in strength.
Thanks
Strength improvements due to software rather than hardware
Moderators: hgm, Rebel, chrisw
-
- Posts: 358
- Joined: Wed Mar 08, 2006 8:36 pm
- Location: UK
-
- Posts: 38
- Joined: Sun Oct 06, 2013 6:16 am
Re: Strength improvements due to software rather than hardwa
All the rating lists are either fixed hardware or at least designed to be equivalent to fixed hardware (and so all rating differences there are from software). Additionally, http://en.chessbase.com/post/komodo-8-t ... -challenge
-
- Posts: 15
- Joined: Sun Dec 07, 2014 3:14 am
- Location: Columbus, Ohio
Re: Strength improvements due to software rather than hardwa
David Levy estimated a 50 to 70 point gain in ELO when processor speed is doubled.
Taking this into account, a typical quality standalone chess "game" from around 1990 ran at around 12 mhz. The best, generally affordable machine played at close to master level. (2200)
If you double the speed, according to Levy, the upgrade should give you 50 to 70 more ELO, or about 2250 -2270 ELO.
Double again to approximately 50 MHz and you get 2300 - 2340.
Go to 100 (typical Pentium, early 90's) and now 2350 - 2410.
200 MHz = 2400 - 2480.
400 MHz = 2450 - 2550.
800 MHz = 2500 - 2620
1600 MHz = 2550 - 2690
So if we average that out, it is somewhere in the low 2600's.
That standalone had at most, a program that was about 1 megabyte.
The average chess instruction in today's chess software (or programmed chess "knowledge") is at least 100 times that large now and likely even greater.
It seems that larger and highly improved programming has added another 600 ELO in the best cases. So the gain seems fairly evenly divided at this stage between software and hardware improvements. (there does seem to be a small lead in the programming end)
If you search the internet for "computer chess" I think you'd be amazed at how much pertinent information is available that addresses your question.
Taking this into account, a typical quality standalone chess "game" from around 1990 ran at around 12 mhz. The best, generally affordable machine played at close to master level. (2200)
If you double the speed, according to Levy, the upgrade should give you 50 to 70 more ELO, or about 2250 -2270 ELO.
Double again to approximately 50 MHz and you get 2300 - 2340.
Go to 100 (typical Pentium, early 90's) and now 2350 - 2410.
200 MHz = 2400 - 2480.
400 MHz = 2450 - 2550.
800 MHz = 2500 - 2620
1600 MHz = 2550 - 2690
So if we average that out, it is somewhere in the low 2600's.
That standalone had at most, a program that was about 1 megabyte.
The average chess instruction in today's chess software (or programmed chess "knowledge") is at least 100 times that large now and likely even greater.
It seems that larger and highly improved programming has added another 600 ELO in the best cases. So the gain seems fairly evenly divided at this stage between software and hardware improvements. (there does seem to be a small lead in the programming end)
If you search the internet for "computer chess" I think you'd be amazed at how much pertinent information is available that addresses your question.
-
- Posts: 12540
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: Strength improvements due to software rather than hardwa
Should be fairly simple to calculate it.James Constance wrote:Hi all
As my A.I. lecturer seems to think that advances in chess engine strength are largely due to improvements in hardware, I was wondering if anyone could point me to somewhere that evaluates the relative contribution of search and evaluation vs hardware to improvements in strength.
Thanks
There are many lists with old programs still in them.
The improvements in both hardware and software are clearly exponential in both cases.
Examine, for instance, Deep Junior on this list. There are many versions tested on the same hardware.
You can also see the improvement from hardware switches.
Code: Select all
THE SSDF RATING LIST 2014-12-11 %135221 games played by 342 computers
Rating + - Games Won Oppo
------ --- --- ----- --- ----
1 Komodo 7.0 MP x64 2GB Q6600 2,4 GHz 3295 41 -36 474 78% 3073
2 Komodo 5.1 MP x64 2GB Q6600 2,4 GHz 3254 28 -26 792 71% 3099
3 Deep Rybka 4 x64 2GB Q6600 2,4 GHz 3209 24 -23 1048 73% 3038
4 Stockfish 3 MP x64 2GB Q6600 2,4 GHz 3207 25 -23 958 70% 3060
5 Deep Hiarcs 14 2GB Q6600 2,4 GHz 3200 23 -22 1020 68% 3071
6 Deep Rybka 3 x64 2GB Q6600 2,4 GHz 3194 22 -21 1371 75% 3001
7 Naum 4.2 MP x64 2GB Q6600 2,4 GHz 3147 21 -21 1083 61% 3071
8 Naum 4 x64 2GB Q6600 2,4 GHz 3119 21 -20 1278 66% 3003
9 Deep Junior Yokoh x64 2GB Q6600 2,4 GHz 3119 30 -30 530 55% 3082
10 Deep Junior 13.3 2GB x64 Q6600 2,4 GHz 3113 23 -22 926 54% 3083
11 Hiarcs 14 256MB Athlon 1200 MHz 3111 42 -38 360 71% 2950
12 Deep Shredder 12 x64 2GB Q6600 2,4 GHz 3103 18 -18 1507 62% 3015
13 Spike 1.4 MP 2GB Q6600 2,4 GHz 3102 19 -19 1349 61% 3024
14 Deep Fritz 13 2GB Q6600 2,4 GHz 3102 24 -24 826 55% 3067
15 Hiarcs 13.1 2GB Q6600 2,4 GHz 3100 24 -24 828 58% 3042
16 Deep Hiarcs 13.2 2GB Q6600 2,4 GHz 3100 27 -26 712 62% 3017
17 Deep Fritz 12 2GB Q6600 2,4 GHz 3092 20 -20 1200 55% 3055
18 Deep Rybka 3 256MB Athlon 1200 MHz 3073 39 -37 332 58% 3017
19 Deep Junior 12 x64 2GB Q6600 2,4 GHz 3071 22 -21 1058 60% 2996
20 Deep Fritz 11 2GB Q6600 2,4 GHz 3060 18 -18 1504 62% 2975
21 Zappa Mexico II x64 2GB Q6600 2,4 GHz 3055 25 -24 816 59% 2989
22 Naum 3.1 x64 2GB Q6600 2,4 GHz 3038 27 -26 692 58% 2984
23 Arasan 17.2 MP x64 2GB Q6600 2,4 GHz 3027 33 -33 422 52% 3011
24 Deep Hiarcs 12 2GB Q6600 2,4 GHz 3017 18 -18 1427 54% 2991
25 Deep Shredder 11 x64 2GB Q6600 2,4 GHz 3012 21 -21 1048 52% 2999
26 Arasan 16.0 MP x64 2GB Q6600 2,4 GHz 3001 33 -34 446 38% 3088
27 Glaurung 2.2 x64 MP 2GB Q6600 2,4 GHz 2993 20 -20 1185 55% 2961
28 Naum 4 256MB Athlon 1200 MHz 2990 26 -26 708 45% 3020
29 Hiarcs 11.2 MP 2GB Q6600 2,4 GHz 2988 20 -20 1125 47% 3006
30 Shredder 12 256MB A1200 MHz 2977 27 -28 632 41% 3038
31 Fritz 13 256MB Athlon 1200 MHz 2976 55 -51 180 63% 2885
32 Deep Junior 10.1 2GB Q6600 2,4 GHz 2966 23 -23 886 46% 2994
33 Fritz 12 256MB A1200 MHz 2948 27 -26 700 62% 2863
34 Rybka 2.3.1 Arena 256MB Athlon 1200 MHz 2920 20 -20 1142 47% 2940
35 Fritz 11 256MB Athlon 1200 MHz 2911 23 -24 853 42% 2966
36 Rybka 1.2 256MB Athlon 1200 MHz 2906 25 -24 933 70% 2760
37 Jonny 4.0 MP 2GB Q6600 2,4 GHz 2900 28 -30 680 29% 3053
38 Deep Fritz 8 2GB Q6600 2,4 GHz 2896 22 -23 1009 36% 2997
39 Deep Shredder 11 256MB Athlon 1200 2895 28 -29 574 45% 2930
40 Shredder 8 MP 2GB Q6600 2,4 GHz 2884 24 -25 908 32% 3010
41 Hiarcs 11.1 256MB Athlon 1200 MHz 2875 21 -21 1077 45% 2911
42 Junior 10.1 256MB Athlon 1200 MHz 2862 17 -18 1585 41% 2922
43 Deep Junior 8 2GB Q6600 2,4 GHz 2860 24 -25 952 30% 3009
44 CM King 3.5 x64 MP 2GB Q6600 2,4 GHz 2855 26 -28 752 29% 3006
45 Fritz 10 256MB Athlon 1200 MHz 2844 34 -32 458 64% 2744
46 Junior 10 256MB Athlon 1200 MHz 2843 22 -22 1042 64% 2740
47 Zap!Chess Zanzibar 256MB Athlon 1200 MH 2834 20 -20 1200 49% 2841
48 Hiarcs 10 HypMod 256MB Athlon 1200 MHz 2834 20 -19 1416 69% 2698
49 Fruit 2.2.1 256MB Athlon 1200 MHz 2829 18 -18 1465 59% 2767
50 Shredder 10 UCI 256MB Athlon 1200 MHz 2824 20 -19 1246 58% 2768
51 Spike 1.2 256MB Athlon 1200 MHz 2819 22 -22 932 49% 2823
52 Fritz 9.0 256MB Athlon 1200 MHz 2814 16 -16 1787 49% 2818
53 Shredder 9.0 UCI 256MB Athlon 1200 Mz 2812 20 -20 1280 66% 2700
54 Pocket Fritz 4 Naum 4.2 Ipaq 214 2805 62 -58 140 61% 2723
55 Shredder 8.0 CB 256MB Athlon 1200 MHz 2800 21 -20 1245 65% 2692
56 Chess Tiger 2007 256MB Athlon 1200 MHz 2796 20 -21 1244 33% 2923
57 Shredder 7.04 UCI 256MB Athlon 1200 MH 2795 21 -20 1269 66% 2682
58 Pro Deo 1.82 256MB Athlon 1200 MHz 2794 33 -34 426 43% 2841
59 Junior 9.0 256MB Athlon 1200 MHz 2783 25 -24 868 63% 2687
60 Deep Fritz 8.0 256MB Athlon 1200 MHz 2780 21 -20 1144 62% 2695
61 Rybka 1.0 beta 128MB K6-2 450 MHz 2774 64 -70 115 38% 2857
62 Junior 8.0 256MB Athlon 1200 MHz 2769 21 -21 1065 55% 2732
63 Revelation Hiarcs 13.3 XScale 500 MHz 2768 63 -58 137 62% 2685
64 Shredder 7.0 256MB Athlon 1200 MHz 2765 26 -25 841 69% 2630
65 Deep Fritz 7.0 256MB Athlon 1200 MHz 2763 24 -23 938 65% 2655
66 Spike 1.1 256MB Athlon 1200 MHz 2761 30 -29 537 57% 2709
67 Pocket Fritz 4 Hiarcs 13 Ipaq 214 2756 57 -54 160 59% 2690
68 Deep Junior 8.0 256MB Athlon 1200 MHz 2754 27 -26 687 61% 2678
69 Fritz 8.0 256MB Athlon 1200 MHz 2754 19 -18 1413 58% 2699
70 Pro Deo 1.86 256MB Athlon 1200 MHz 2750 57 -72 190 19% 2993
71 Zap!Chess 256MB Athlon 1200 MHz 2743 26 -26 682 48% 2761
72 Hiarcs 9.0 256MB Athlon 1200 MHz 2738 17 -18 1522 48% 2752
73 Fritz 7.0 256MB Athlon 1200 MHz 2737 27 -27 674 61% 2660
74 Gandalf 6.0 256MB Athlon 1200 MHz 2735 23 -22 935 54% 2704
75 Chess Tiger 2004 256MB Athlon 1200 MH 2724 21 -21 1088 52% 2708
76 Gambit Tiger 2.0 256MB Athlon 1200 2722 26 -26 683 55% 2689
77 Shredder 6.0 Pad UCI 256MB Athlon 1200 2720 22 -21 1073 60% 2647
78 Chess Tiger 15.0 256MB Athlon 1200 MH 2718 19 -19 1299 54% 2693
79 Pocket Fritz 3 Hiarcs Ipaq 214 624 MHz 2717 50 -48 200 58% 2663
80 Chess Tiger 14.0 CB 256MB Athlon 1200 2714 28 -27 637 58% 2661
81 Shredder 6.0 256MB Athlon 1200 MHz 2713 29 -28 587 60% 2644
82 Pro Deo 1.1 256MB Athlon 1200 MHz 2711 24 -23 876 57% 2659
83 Deep Fritz 256MB Athlon 1200 MHz 2710 24 -24 791 51% 2702
84 Chessmaster 9000 256MB Athlon 1200 MHz 2706 35 -34 396 54% 2677
85 Revelation Shredder 12 XScale 500 MHz 2705 60 -58 140 56% 2665
86 Pocket Shredder Ipaq 114 624 MHz 2704 51 -48 200 61% 2628
87 Junior 7.0 256MB Athlon 1200 MHz 2697 21 -21 1048 51% 2690
88 Shredder 7.0 UCI 128MB K6-2 450 MHz 2690 24 -24 834 43% 2741
89 Rebel 12.0 256MB Athlon 1200 MHz 2684 29 -29 566 51% 2679
90 Hiarcs 8.0 256MB Athlon 1200 MHz 2680 20 -20 1116 51% 2673
91 Ruffian 1.0.1 256MB Athlon 1200 MHz 2676 23 -23 896 48% 2693
92 Rebel Century 4.0 256MB Athlon 1200 2673 29 -28 590 60% 2604
93 Deep Sjeng 1.5a 256MB Athlon 1200 MHz 2672 31 -31 493 52% 2661
94 Chess Tiger 15.0 128MB K6-2 450 MHz 2670 24 -25 834 40% 2743
95 Deep Fritz 7.0 128MB K6-2 450 MHz 2663 21 -21 1072 42% 2717
96 Shredder 5.32 256MB Athlon 1200 MHz 2662 24 -24 819 49% 2669
97 Gandalf 4.32h 256MB Athlon 1200 MHz 2661 28 -28 604 52% 2650
98 CEBoard Fruit 2.3.1 XScale 400 400 MHz 2658 47 -47 215 53% 2639
99 Gandalf 5.0 256MB Athlon 1200 MHz 2653 36 -36 364 45% 2689
100 Hiarcs Palm Chess 12.1 T5 2653 63 -64 120 48% 2665
101 Deep Fritz 128MB K6-2 450 MHz 2646 20 -20 1151 56% 2606
102 Gambit Tiger 2.0 128MB K6-2 450 MHz 2643 24 -24 835 58% 2587
103 Chess Tiger 14.0 CB 128MB K6-2 450 MHz 2639 22 -22 973 54% 2614
104 Gandalf 5.1 256MB Athlon 1200 MHz 2637 25 -25 758 55% 2604
105 Fritz 7.0 128MB K6-2 450 MHz 2634 36 -35 368 54% 2605
106 Ruffian 2.0.0 256MB Athlon 1200 MHz 2631 48 -49 205 46% 2660
107 Junior 7.0 128MB K6-2 450 MHz 2631 20 -20 1200 56% 2588
108 Revelation Rybka 2.2 XScale 500 MHz 2628 47 -44 240 62% 2545
109 Shredder 6.0 UCI 128MB K6-2 450 MHz 2617 42 -42 264 52% 2607
110 Crafty 18.12/CB 256MB Athlon 1200 MH 2614 20 -21 1141 39% 2690
111 Fritz 6.0 128MB K6-2 450 MHz 2613 16 -16 1751 48% 2626
112 Gromit 3.11.9 256MB Athlon 1200 MHz 2610 43 -45 246 43% 2661
113 Shredder 5.32 128MB K6-2 450 MHz 2608 19 -20 1242 42% 2667
114 Yace Paderborn 256MB Athlon 1200 MH 2607 35 -35 389 48% 2620
115 Chess Tiger 12.0 DOS 128MB K6-2 450 2599 18 -18 1448 48% 2612
116 Junior 6.0 128MB K6-2 450 MHz 2595 13 -13 2677 43% 2646
117 Hiarcs 8.0 128MB K6-2 450 MHz 2582 31 -33 480 39% 2663
118 Shredder 5.0 128MB K6-2 450 MHz 2582 16 -16 1818 44% 2623
119 Shredder 4.0 128MB K6-2 450 MHz 2571 22 -22 986 58% 2512
120 Rebel Century 4.0 128MB K6-2 450 MHz 2568 51 -52 178 46% 2594
121 Nimzo 7.32 128MB K6-2 450 MHz 2558 17 -18 1542 45% 2596
122 Fritz 5.32 128MB K6-2 450 MHz 2556 20 -20 1194 48% 2573
123 Nimzo 8.0 128MB K6-2 450 MHz 2553 19 -19 1362 43% 2602
124 Junior 5.0 128MB K6-2 450 MHz 2542 20 -20 1108 50% 2543
125 Hiarcs 7.32 128MB K6-2 450 MHz 2538 17 -17 1736 38% 2624
126 Gandalf 5.0 128MB K6-2 450 MHz 2537 66 -62 122 59% 2474
127 Gandalf 4.32f 128MB K6-2 450 MHz 2535 27 -27 627 51% 2528
128 Crafty 19.17 256MB Athlon 1200 MH 2531 41 -45 304 30% 2680
129 Gandalf 4.32h 128MB K6-2 450 MHz 2526 34 -33 418 52% 2511
130 Hiarcs 7.01 128MB K6-2 450 MHz 2523 30 -31 525 43% 2574
131 SOS 128MB K6-2 450 MHz 2522 14 -14 2556 33% 2642
132 Rebel Century 3.0 128MB K6-2 450 MHz 2521 29 -29 546 49% 2527
133 Chessmaster 8000 128MB K6-2 450 MHz 2520 43 -44 251 45% 2552
134 Pocket Fritz 2 XScale 400 MHz 2517 41 -41 283 51% 2509
135 Goliath Light 128MB K6-2 450 MHz 2506 17 -18 1961 27% 2676
136 Pocket Fritz 3 Glaurung 2.1 Ipaq 614C 2501 59 -67 140 33% 2628
137 Fritz 5.32 64MB P200 MMX 2499 13 -13 3023 38% 2588
138 Crafty 17.07/CB 128MB K6-2 450 MHz 2499 19 -19 1459 33% 2621
139 Crafty 18.12/CB 128MB K6-2 450 MHz 2491 25 -27 929 25% 2682
140 Nimzo 99 128MB K6-2 450 MHz 2486 21 -22 1016 44% 2527
141 Resurrection Rybka 2.2 StrongARM 203 MH 2484 43 -42 260 51% 2477
142 MChess Pro 8.0 128MB K6-2 450 MHz 2481 25 -26 753 40% 2551
143 Genius 6.5 128MB K6-2 450 MHz 2479 28 -28 605 48% 2492
144 Chessmaster 6000 64MB P200 MMX 2477 65 -54 184 76% 2282
145 Hiarcs 7.32 64MB P200 MMX 2473 20 -20 1134 52% 2460
146 Hiarcs 7.0 64MB P200 MMX 2463 21 -20 1112 55% 2425
147 Fritz 5.0 PB29% 67MB P200 MMX 2462 23 -22 1005 66% 2347
148 Nimzo 99 64MB P200 MMX 2452 23 -23 885 51% 2445
149 Junior 5.0 64MB P200 MMX 2437 18 -18 1458 45% 2475
150 Nimzo 98 58MB P200 MMX 2433 19 -19 1352 51% 2429
151 Rebel 9.0 47MB P200 MMX 2422 23 -23 920 61% 2342
152 Hiarcs 6.0 49MB P200 MMX 2421 24 -24 829 56% 2378
153 Shredder 3.0 64MB P200 MMX 2421 19 -19 1262 48% 2433
154 Rebel 8.0 51MB P200 MMX 2412 21 -22 1011 47% 2436
155 Zarkov 5.0 64MB P200 MMX 2412 24 -25 789 40% 2481
156 MChess Pro 6.0 41MB P200 MMX 2410 24 -24 831 52% 2397
157 Shredder 2.0 58MB P200 MMX 2405 19 -19 1314 44% 2444
158 MChess Pro 8.0 64MB P200 MMX 2398 26 -25 718 53% 2376
159 MChess Pro 7.1 46MB P200 MMX 2397 21 -21 1042 53% 2375
160 Genius 5.0 DOS 46MB P200 MMX 2395 20 -20 1197 49% 2399
161 Resurrection Fruit '05 StrongARM 203 MH 2393 67 -62 120 60% 2321
162 Hiarcs 9.5a/9.6 Palm TungstenE OMAP 126 2392 34 -35 400 45% 2426
163 Chess Tiger 11.8 Pentium 90 MHz 2386 42 -42 261 50% 2388
164 CEBoard Crafty 2004 HP RX4240 400 MHz 2372 48 -50 200 43% 2423
165 Gandalf 3.0 64MB P200 MMX 2367 41 -39 307 59% 2300
166 WChess2000 64MB P200 MMX 2362 30 -31 542 37% 2453
167 Kallisto II 64MB P200 MMX 2346 34 -34 403 52% 2331
168 Rebel 9.0 Pentium 90 MHz 2340 22 -22 930 47% 2359
169 Rebel 8.0 Pentium 90 MHz 2339 18 -18 1435 55% 2305
170 Hiarcs 6.0 Pentium 90 MHz 2336 18 -18 1437 51% 2332
171 Genius 5.0 DOS Pentium 90 MHz 2333 17 -17 1558 47% 2352
172 Hiarcs 5.0 Pentium 90 MHz 2328 37 -36 363 59% 2264
173 MChess Pro 6.0 Pentium 90 MHz 2314 16 -17 1726 45% 2347
174 Genius 3.0 Pentium 90 MHz 2311 24 -24 845 60% 2238
175 Rebel 6.0 Pentium 90 MHz 2306 31 -29 540 60% 2235
176 MChess Pro 5.0 Pentium 90 MHz 2306 26 -25 769 63% 2215
177 Genius 4.0 DOS Pentium 90 MHz 2303 23 -22 944 60% 2235
178 Nimzo 3.5 Pentium 90 MHz 2297 22 -22 998 46% 2326
179 Rebel 7.0 Pentium 90 MHz 2297 25 -24 799 59% 2231
180 Junior 4.0 Pentium 90 MHz 2292 21 -22 1035 42% 2345
181 Hiarcs 4.0 Pentium 90 MHz 2291 25 -24 786 54% 2262
182 Shredder 1.0 Pentium 90 MHz 2290 54 -54 165 50% 2292
183 Chessmaster 5000 Pentium 90 MHz 2286 49 -45 240 67% 2163
184 Nimzo 3.0 Pentium 90 MHz 2281 25 -24 787 58% 2222
185 Hiarcs 3.0 Pentium 90 MHz 2273 30 -29 545 56% 2229
186 R30 v. 2.5 2271 42 -38 343 69% 2134
187 Genius 4.0 DOS 486/50-66 MHz 2267 23 -23 900 49% 2271
188 CometA90 64MB P200 MMX 2261 35 -37 378 38% 2344
189 Junior 3.3-3.5 Pentium 90 MHz 2260 30 -30 516 47% 2279
190 Chess Genius 3 Iphone 4 ARM A8 800MHz 2258 82 -69 100 69% 2119
191 MB P200 MMX 2255 42 -43 268 43% 2304
192 MChess Pro 4.0 Pentium 90 MHz 2255 29 -28 578 52% 2239
193 Genius 3.0 486/50-66 MHz 2254 24 -23 910 62% 2168
194 Fritz 3.0 Pentium 90 MHz 2245 18 -18 1478 41% 2305
195 WChess 1.06 Pentium 90 MHz 2235 20 -20 1222 39% 2311
196 Mephisto Genius 2.0 486/50-66 MHz 2235 23 -23 917 58% 2177
197 Fritz 4.0 Pentium 90 MHz 2233 40 -38 324 60% 2164
198 Palm Tiger 2009 Tung C 400 MHz 2230 51 -53 180 43% 2277
199 Kallisto 1.98 Pentium 90 MHz 2229 21 -22 1049 42% 2286
200 MChess Pro 5.0 486/50-66 MHz 2229 26 -26 691 50% 2231
201 Rebel 7.0 486/50-66 MHz 2227 24 -25 777 49% 2235
202 MChess Pro 4.0 486/50-66 MHz 2209 24 -23 846 53% 2186
203 Chess Machine 30-32 MHz Schroeder 3.1 2208 28 -27 646 63% 2119
204 WChess 1.03 486/50-66 MHz 2199 26 -26 677 49% 2205
205 Hiarcs 3.0 486/50-66 MHz 2197 23 -23 865 50% 2194
206 Rebel 6.0 486/50-66 MHz 2197 23 -23 887 55% 2163
207 Ch.Machine 30 MHz King 2.0 aggr/R30off 2195 20 -20 1239 63% 2100
208 Meph Genius 68 030 33 MHz 2193 41 -40 287 53% 2173
209 Mephisto London 68 030 33 MHz 2190 31 -31 482 50% 2188
210 Chessmaster 4000 486/50-66 MHz 2190 35 -33 462 66% 2072
211 Chess Genius 1.0 486/50-66 MHz 2183 22 -22 971 55% 2148
212 Rebel Decade Pentium 90 MHz 2176 29 -29 563 45% 2213
213 Chess Machine 30 MHz Schroeder 3.0 2174 65 -55 168 73% 2002
214 Mephisto Gideon Pro 486/50-66 MHz 2170 35 -34 421 62% 2084
215 MChess Pro 3.5 486/50-66 MHz 2168 22 -22 958 51% 2160
216 MChess Pro 3.12 486/50-66 MHz 2164 37 -34 438 69% 2023
217 Fritz 3.0 486/50-66 MHz 2160 19 -19 1246 46% 2191
218 Mephisto Lyon 68030 36 MHz 2151 28 -27 716 68% 2017
219 Chess Genius 1.4 SX1 OMAP 310 120 MHz 2148 50 -47 210 60% 2080
220 Chess Genius 1.0 486/33 MHz 2139 31 -31 483 50% 2142
221 Mephisto Portorose 68030 36 MHz 2139 43 -38 407 77% 1929
222 Mephisto Vancouver 68030 36 MHz 2127 36 -33 471 71% 1972
223 Meph RISC 2 1 MB 2127 61 -66 125 39% 2206
224 Kasparov SPARC 20 MHz 2125 23 -23 912 45% 2158
225 MChess Pro 3.12 486/33 MHz 2125 50 -47 208 60% 2052
226 Berlin Pro 68 020 24 MHz 2125 24 -23 850 58% 2071
227 Kallisto 1.82-1.83 486/50-66 MHz 2109 20 -21 1117 42% 2166
228 Chess Tiger 14.9 Palm m515 16MB 42MHz 2104 68 -75 100 39% 2182
229 Hiarcs Master 2.0 486/33 MHz 2103 44 -43 249 53% 2085
230 Mephisto RISC 1 MB ARM 2 14 MHz 2102 19 -19 1356 55% 2068
231 Comet32 Pentium 90 MHz 2100 24 -26 1011 24% 2304
232 Mephisto Montreux ARM 14 MHz 512K 2095 27 -25 809 68% 1961
233 MChess 1.1-1.71 486/33 2093 46 -41 326 74% 1915
234 Saitek RISC 2500 ARM2 14 MHz 128K 2092 19 -19 1276 54% 2066
235 Chess Machine Schroeder 512K ARM2 16MHz 2091 27 -26 703 62% 2009
236 Atlanta SH7000 20 MHz 2083 28 -27 707 67% 1964
237 Chess Machine The King 512K ARM2 16MHz 2073 32 -31 473 55% 2036
238 Mephisto Vancouver 68020 12 MHz 2060 23 -22 1041 65% 1951
239 Mephisto Lyon 68020 12 MHz 2049 18 -18 1442 59% 1987
240 Socrates 3.0 486/33 MHz 2040 48 -49 203 47% 2061
241 Fritz 2.0 486/33 MHz 2030 29 -30 547 44% 2069
242 Mephisto Portorose 68020 12 MHz 2028 27 -25 896 73% 1857
243 MChess 1.1-1.71 386/25-33 MHz 2025 37 -34 408 65% 1918
244 Star Sapphire 2024 35 -35 389 52% 2014
245 Fidelity Elite 68030 32 MHz (vers.9) 2021 31 -30 521 60% 1949
246 Mephisto Berlin 68 000 12 MHz 2014 22 -22 986 57% 1967
247 Sapphire II 2008 33 -31 484 62% 1924
248 Mephisto Vancouver 68000 12 MHz 2003 20 -20 1159 54% 1977
249 Mephisto Lyon 68000 12 MHz 2002 20 -20 1190 56% 1960
250 Mephisto MM 4 Turbo Kit 6502 16 MHz 1994 48 -43 292 73% 1820
251 Novag Sapphire H8 10 MHz 1990 21 -21 1074 46% 2015
252 Atlanta Brute Force SH7000 20 MHz 1981 52 -50 185 55% 1946
253 Milano Pro SH7000 20 MHz 1976 32 -30 509 60% 1904
254 Fidelity Mach IV 68020 20 MHz 1975 18 -18 1495 48% 1989
255 Mephisto Almeria 68020 12 MHz 1973 42 -39 305 62% 1887
256 Hiarcs Master 1.0 486/33 MHz 1971 47 -47 214 48% 1987
257 Mephisto Portorose 68000 12 MHz 1945 24 -23 867 60% 1872
258 Mephisto Polgar 6502 10 MHz 1942 42 -41 283 59% 1880
259 Fritz 1.0 486/33 MHz 1941 48 -46 215 55% 1909
260 Gandalf 2.1 Pentium 90 MHz 1939 47 -53 242 27% 2113
261 Fidelity Elite 68000 x 2 (vers. 5) 1936 49 -47 209 57% 1888
262 Nimzo 2.2.1 486/33 MHz 1936 45 -47 229 42% 1990
263 Zarkov 3.0 486/25-33 MHz 1931 45 -48 232 39% 2009
264 Mephisto Roma 68020 14 MHz 1930 28 -26 712 65% 1821
265 Rex Chess 2.3 386/25-33 MHz 1929 65 -61 126 59% 1865
266 Mephisto Dallas 68020 12 MHz 1924 36 -33 492 72% 1763
267 Mephisto Almeria 68000 12 MHz 1920 27 -26 694 59% 1856
268 Zarkov 2.5 386/25-33 MHz 1919 56 -53 168 61% 1842
269 Kasparov Brute Force H8 10 MHz 1917 22 -22 966 43% 1967
270 Fritz 1.0 386/25-33 MHz 1908 64 -66 113 46% 1939
271 Novag Diablo 68000 16 MHz 1907 20 -21 1163 41% 1974
272 Fidelity Mach III 68000 16 MHz 1894 14 -14 2410 52% 1881
273 Complete Chess System 486/33 MHz 1886 46 -47 221 47% 1908
274 Kasparov President/GK-2100 H8 10 MHz 1880 28 -29 581 46% 1907
275 Mephisto MM 5 6502 5 MHz 1876 18 -18 1396 46% 1902
276 Chess Genius 1.5 Palm m515 16MB 42MHz 1872 69 -81 100 32% 2004
277 Mephisto Polgar 6502 5 MHz 1872 16 -16 1838 40% 1940
278 Mephisto Dallas 68000 12 MHz 1871 21 -20 1177 61% 1796
279 Mephisto Roma 68000 12 MHz 1870 18 -18 1519 56% 1829
280 Mephisto Milano 6502 5 MHz 1866 22 -23 950 39% 1941
281 Novag Super Expert C 6502 6 MHz Sel 5 1861 17 -17 1555 46% 1892
282 Zarkov 2.6 386/25-33 MHz 1860 64 -61 125 56% 1819
283 Mephisto Academy 6502 5 MHz 1841 17 -17 1595 43% 1887
284 Chessmaster 3000 386/25-33 MHz 1834 67 -75 103 36% 1931
285 Rex Chess 2.3 386/16-20 MHz 1831 52 -52 174 49% 1838
286 Saitek Leonardo Maestro B 6502 18 MHz 1828 73 -66 105 62% 1744
287 Mephisto Amsterdam 68000 12 MHz 1827 22 -21 1016 57% 1775
288 Novag Super Expert B 6502 6 MHz sel 4 1821 29 -29 564 48% 1837
289 Mephisto Mega IV 6502 4.9 MHz 1817 17 -17 1604 43% 1863
290 Saitek Maestro D 6502 10 MHz 1817 25 -25 736 51% 1811
291 Fidelity Excel Mach IIc 68000 12 MHz 1817 22 -22 975 50% 1815
292 Novag Expert Turbo Kit 6502 16 MHz 1811 68 -62 120 61% 1731
293 Mephisto MM 4 6502 5 MHz 1806 19 -19 1274 50% 1805
294 Mephisto Modena 6502 4 MHz 1799 27 -28 655 40% 1869
295 Kasparov GK-2000 H8 10 MHz 1799 28 -29 593 42% 1854
296 Fidelity Travel Master H8/330 10 MHz 1796 67 -86 123 22% 2013
297 Psion Atari 68000 8 MHz 1784 18 -18 1483 43% 1830
298 CXG Sphinx Galaxy 6502 4 MHz 1783 17 -17 1733 37% 1875
299 Novag Ruby H8 10 MHz 1781 29 -30 545 42% 1836
300 Fidelity Excel Mach IIa 68000 12 MHz 1780 45 -46 226 47% 1800
301 Saitek Turboking II 6502 5 MHz 1771 22 -23 963 37% 1865
302 Conchess Plymate Victoria 6502 5.5 MHz 1770 26 -27 701 40% 1842
303 Fidelity Excel Club 68000 12 MHz 1759 22 -22 931 51% 1751
304 Mephisto Mega IV Brute Force 1751 34 -35 422 38% 1840
305 Novag Super Expert/Forte A 6502 5 MHz 1742 18 -19 1425 37% 1835
306 Fidelity Par Excellence 6502 5 MHz 1738 21 -21 1014 48% 1751
307 Fidelity Avant Garde 6502 5 MHz 1737 20 -20 1196 46% 1768
308 Mephisto Rebell 6502 5 MHz 1726 19 -19 1391 38% 1811
309 Saitek Stratos/Analys B 6502 6 MHz 1723 20 -20 1213 38% 1809
310 Mephisto Super Mondial 6502 4 MHz 1722 18 -19 1455 35% 1828
311 Novag Forte B 6502 5 MHz 1719 24 -24 837 44% 1762
312 Saitek Leonardo Maestro A 6502 6 MHz 1714 29 -30 547 45% 1748
313 Novag Forte A 6502 5 MHz 1711 22 -22 964 48% 1724
314 Conchess Plymate 6502 5.5 MHz 1711 15 -16 2040 40% 1783
315 Saitek Simultano/Corona C 6502 5 MHz 1709 23 -24 969 33% 1834
316 Fidelity Excellence 6502 4 MHz 1704 17 -18 1562 43% 1754
317 Novag Expert 6502 4 MHz 1690 24 -24 841 42% 1749
318 Fidelity Elegance 6502 3.6 MHz 1681 40 -39 303 54% 1653
319 Conchess Plymate 6502 4 MHz 1681 38 -37 333 55% 1650
320 Mephisto MM 2 6502 3.7 MHz 1674 48 -47 208 52% 1658
321 Saitek Turbostar 432 6502 4 MHz 1666 22 -22 975 44% 1708
322 Fidelity Excellence 6502 3 MHz 1660 22 -23 951 42% 1718
323 Saitek Kasparov Blitz 1638 44 -47 239 37% 1728
324 Novag Super Nova 1637 37 -38 350 40% 1704
325 Novag Super Constellation 6502 4 MHz 1633 18 -18 1595 34% 1745
326 Conchess Glasgow 6502 4 MHz 1621 38 -38 327 50% 1623
327 Chessplayer 2150 Atari/Amiga 1618 57 -64 146 33% 1740
328 The Final Chesscard 6502 5 MHz 1602 57 -68 156 27% 1773
329 Novag Super VIP 1591 53 -59 174 32% 1719
330 Chessmaster 2100 Amiga 68000 1589 66 -81 120 25% 1776
331 Mephisto Europa 1585 49 -52 190 42% 1643
332 Saitek Superstar 36K 6502 2 MHz 1582 24 -25 914 30% 1733
333 Fidelity Elite A/S 6502 3.2 MHz 1580 23 -25 1023 27% 1750
334 Chess Champion 2175 Atari/Amiga 1577 56 -65 157 30% 1725
335 Conchess Glasgow 6502 2 MHz 1571 25 -26 829 33% 1692
336 Novag Quattro 6502 4 MHz 1566 30 -32 560 33% 1693
337 Fidelity Prestige 6502 4 MHz 1561 63 -66 118 44% 1603
338 Chessmaster 2000 Atari 68000 8 MHz 1556 42 -48 312 25% 1743
339 Novag Constellation 6502 3.6 MHz 1548 33 -34 448 37% 1643
340 Novag Constellation Primo 1541 56 -58 149 44% 1586
341 Novag Constellation 6502 2 MHz 1490 34 -36 439 32% 1625
342 CXG Super Enterprise 1466 38 -42 376 28% 1635
Here is the SSDF list of champions:
Code: Select all
Rating list year-end leaders Year Program Hardware Rating
1984 Novag Super Constellation 6502 4 MHz 1631
1985 Mephisto Amsterdam 68000 12 MHz 1827
1986 Mephisto Amsterdam 68000 12 MHz 1827
1987 Mephisto Dallas 68020 14 MHz 1923
1988 Mephisto MM 4 Turbo Kit 6502 16 MHz 1993
1989 Mephisto Portorose 68020 12 MHz 2027
1990 Mephisto Portorose 68030 36 MHz 2138
1991 Mephisto Vancouver 68030 36 MHz 2127
1992 Chess Machine Schröder 3.0 ARM2 30 MHz 2174
1993 Mephisto Genius 2.0 486/50-66 MHz 2235
1995 MChess Pro 5.0 Pentium 90 MHz 2306
1996 Rebel 8.0 Pentium 90 MHz 2337
1997 HIARCS 6.0 49MB P200 MMX 2418
1998 Fritz 5.0 PB29% 67MB P200 MMX 2460
1999 Chess Tiger 12.0 DOS 128MB K6-2 450 MHz 2594
2000 Fritz 6.0 128MB K6-2 450 MHz 2607
2001 Chess Tiger 14.0 CB 256MB Athlon 1200 2709
2002 Deep Fritz 7.0 256MB Athlon 1200 MHz 2759
2003 Shredder 7.04 UCI 256MB Athlon 1200 MHz 2791
2004 Shredder 8.0 CB 256MB Athlon 1200 MHz 2800
2005 Shredder 9.0 UCI 256MB Athlon 1200 MHz 2808
2006 Rybka 1.2 256MB Athlon 1200 MHz 2902
2007 Rybka 2.3.1 Arena 256MB Athlon 1200 MHz 2935
2008 Deep Rybka 3 2GB Q6600 2.4 GHz 3238
2009 Deep Rybka 3 2GB Q6600 2.4 GHz 3232
2010 Deep Rybka 3 2GB Q6600 2.4 GHz 3227
2011 Deep Rybka 4 2GB Q6600 2.4 GHz 3216
2012 Deep Rybka 4 x64 2GB Q6600 2.4 GHz 3221
2013 Komodo 5.1 MP x64 2GB Q6600 2.4 GHz 3241
2014 Komodo 7.0 MP x64 2GB Q6600 2.4 GHz 3295
Since those programs are still around, you can test it for yourselves.
Consider also lists like CCRL and CEGT. These lists actually use calibrated compute power so that every engine has the same horsepower (as a function of threads, of course) so that single thread programs will have the same power devoted to them whenever they are tested.
Fritz has gained 328 Elo from Fritz 8 to Fritz 14 ON THE SAME Exact hardware. Considering the exponential nature of the Elo scale, that is an enormous increase. I have not done the math, but my guess is that it will be commensurate with the gains from hardware. See:
http://www.computerchess.org.uk/ccrl/40 ... +opponents
-
- Posts: 10948
- Joined: Wed Jul 26, 2006 10:21 pm
- Full name: Kai Laskos
Re: Strength improvements due to software rather than hardwa
At least since year 2003, definitely more improvement in software than hardware.James Constance wrote:Hi all
As my A.I. lecturer seems to think that advances in chess engine strength are largely due to improvements in hardware, I was wondering if anyone could point me to somewhere that evaluates the relative contribution of search and evaluation vs hardware to improvements in strength.
Thanks
-
- Posts: 1221
- Joined: Wed Mar 08, 2006 8:28 pm
- Location: Florida, USA
Re: Strength improvements due to software rather than hardwa
Back in the 1990's the "it's-all-down-to-improvements-in-the-hardware" mantra was not too far from the truth.
However, since 2004 the search algorithm improvements have been huge. This is one reason why a modern iPhone running Stockfish would slaughter Deep Blue.
Steve
However, since 2004 the search algorithm improvements have been huge. This is one reason why a modern iPhone running Stockfish would slaughter Deep Blue.
Steve
http://www.chessprogramming.net - Maverick Chess Engine
-
- Posts: 358
- Joined: Wed Mar 08, 2006 8:36 pm
- Location: UK
Re: Strength improvements due to software rather than hardwa
Thanks all for your replies. You've certainly pointed me in the right direction to make some calculations myself.
I'd still be interested in any other articles that address the subject, if you come across any.
Thanks!
I'd still be interested in any other articles that address the subject, if you come across any.
Thanks!
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: Strength improvements due to software rather than hardwa
Just read the thread:James Constance wrote:Thanks all for your replies. You've certainly pointed me in the right direction to make some calculations myself.
I'd still be interested in any other articles that address the subject, if you come across any.
Thanks!
http://talkchess.com/forum/viewtopic.php?p=369240
-
- Posts: 918
- Joined: Mon Jan 05, 2009 7:40 pm
- Location: Germany
- Full name: Engin Üstün
Re: Strength improvements due to software rather than hardwa
agree, this is my plan too, i want to make the software better not to stick only on the speed of the hardware to become more strength.