Author: Robert Henry Durrett
Date: 14:17:04 06/13/02
Go up one level in this thread
On June 13, 2002 at 15:26:17, pavel wrote: >On June 13, 2002 at 14:51:46, Robert Henry Durrett wrote: > >>On June 13, 2002 at 13:32:20, Russell Reagan wrote: >> >>>On June 13, 2002 at 12:54:16, Robert Henry Durrett wrote: >>> >>>>I interpret this to imply that "everybody does it." >>>> >>>>That leaves the question: How much higher are the SSDF ratings than they would >>>>have been if commercial versions [Assumed to be not optimized for one specific >>>>computer] were used? >>>> >>>>Not really asking about books. More interested in knowing how much difference >>>>optimization for a specific computer would make. >>>> >>>>Bob D. >>> >>>Not much. You're talking about a very small gain in speed. The top programs are >>>already probably optimized out the wazoo, and the small gain that optimizing for >>>a particular processor will bring translates to even smaller gains in ELO. If >>>you get a few more NPS, then maybe you'll find a "better" (or at least >>>different) move maybe once during the game, and that's probably not going to >>>translate into turning a loss into a draw or turning a draw into a win, IMO. >>> >>>Russell >> >>Well, I was prompted to think about this topic by the statement: "Hiarcs 8 was >>NOT made for slow computer such as an AMD 450 Mhz as the SSDF decided to test it >>against Nimzo 8." made in another thread. >> >>I realize that big programs require big computers. That is clear. >> >>But if all the chess engines were to be tested on the same computers, you would >>have thought that all of the chess engine developers would have optimized their >>programs for the computers they were to run on. This includes "paring down" a >>big program if it will have to run on a "small" computer. That's what got me to >>thinking, anyway. I couldn't understand how Hiarcs 8 had been zapped. >> >>For whatever that's worth. >> >>Bob D. >> >>Bob D. > > >SSDF tests allmost all TOP programs in both type of processors 450mhz and >1200mhz, the idea is to see how much strength a program gets from faster >processor. >It has been done AFAIK with fritz7, crafty, junior7, chesstiger14, just to name >a few and almost most other top programs. > >as you can see... > > > Rating + - Games Won Av.opp >1 Fritz 7.0 256MB Athlon 1200 MHz 2730 33 -31 494 64% 2626 >2 Chess Tiger 14.0 CB 256MB Athlon 1200 2722 33 -32 477 63% 2626 >3 Gambit Tiger 2.0 256MB Athlon 1200 2720 34 -33 441 62% 2635 >4 Deep Fritz 256MB Athlon 1200 MHz 2714 33 -32 482 63% 2623 >5 Shredder 6.0 256MB Athlon 1200 MHz 2713 35 -34 432 64% 2611 >6 Junior 7.0 256MB Athlon 1200 MHz 2693 31 -31 511 57% 2641 >7 Rebel Century 4.0 256MB Athlon 1200 MHz 2684 34 -32 470 64% 2585 >8 Shredder 5.32 256MB Athlon 1200 MHz 2670 31 -30 536 56% 2624 >9 Gandalf 4.32h 256MB Athlon 1200 MHz 2653 34 -33 430 54% 2625 >10 Deep Fritz 128MB K6-2 450 MHz 2652 23 -23 945 61% 2570 >11 Gandalf 5.1 256MB Athlon 1200 MHz 2646 29 -28 595 57% 2594 >12 Gandalf 5.0 256MB Athlon 1200 MHz 2642 49 -50 202 46% 2673 >13 Gambit Tiger 2.0 128MB K6-2 450 MHz 2640 30 -29 592 66% 2521 >14 Fritz 7.0 128MB K6-2 450 MHz 2631 45 -44 250 56% 2592 >14 Junior 7.0 128MB K6-2 450 MHz 2631 27 -26 739 67% 2507 >16 Chess Tiger 14.0 CB 128MB K6-2 450 MHz 2630 28 -27 652 62% 2541 >16 Shredder 6.0 UCI 128MB K6-2 450 MHz 2630 65 -62 124 57% 2578 >18 Fritz 6.0 128MB K6-2 450 MHz 2623 23 -22 1014 63% 2529 >19 Crafty 18.12/CB 256MB Athlon 1200 MHz 2617 31 -30 519 53% 2595 >20 Shredder 5.32 128MB K6-2 450 MHz 2606 31 -30 545 62% 2521 >21 Junior 6.0 128MB K6-2 450 MHz 2588 18 -18 1483 58% 2533 >22 Chess Tiger 12.0 DOS 128MB K6-2 450 MHz 2587 22 -22 1021 55% 2551 >23 Shredder 5.0 128MB K6-2 450 MHz 2586 20 -20 1176 56% 2544 >24 Rebel Century 4.0 128MB K6-2 450 MHz 2571 59 -61 138 44% 2612 >25 Shredder 4.0 128MB K6-2 450 MHz 2568 23 -22 986 58% 2508 >26 Nimzo 8.0 128MB K6-2 450 MHz 2554 24 -24 846 53% 2536 >27 Fritz 5.32 128MB K6-2 450 MHz 2553 24 -23 890 55% 2514 >28 Nimzo 7.32 128MB K6-2 450 MHz 2551 21 -21 1070 54% 2520 >29 Junior 5.0 128MB K6-2 450 MHz 2537 22 -22 996 52% 2520 >30 Gandalf 5.0 128MB K6-2 450 MHz 2532 73 -68 102 60% 2458 >31 Gandalf 4.32f 128MB K6-2 450 MHz 2531 28 -28 627 51% 2524 >32 Hiarcs 7.32 128MB K6-2 450 MHz 2527 21 -21 1131 47% 2546 >33 Hiarcs 7.01 128MB K6-2 450 MHz 2521 30 -31 525 43% 2573 >34 SOS 128MB K6-2 450 MHz 2520 19 -19 1307 47% 2539 >35 Gandalf 4.32h 128MB K6-2 450 MHz 2518 36 -36 378 53% 2498 >36 Rebel Century 3.0 128MB K6-2 450 MHz 2517 30 -30 546 49% 2523 >37 Chessmaster 8000 128MB K6-2 450 MHz 2516 44 -45 251 45% 2549 >38 Goliath Light 128MB K6-2 450 MHz 2507 24 -24 857 42% 2565 >39 Crafty 17.07/CB 128MB K6-2 450 MHz 2490 23 -23 912 47% 2513 >40 Nimzo 99 128MB K6-2 450 MHz 2485 22 -22 996 44% 2528 >41 MChess Pro 8.0 128MB K6-2 450 MHz 2478 25 -26 753 40% 2549 >42 Genius 6.5 128MB K6-2 450 MHz 2474 29 -29 565 48% 2487 > >http://w1.859.telia.com/~u85924109/ssdf/list.htm > A VERY impressive table!!! > >"Hiarcs 8 was NOT made for slow computer such as an AMD 450 Mhz as the SSDF >decided to test it against Nimzo 8." > >so the statement is moot, when you know why they did, what they did. > >Also statements like Hiarcs 8 is better in long(er) time control than relatively >short time control doesnt make sense to me. > >I dont think SSDF is the last thing on earth for all programmers. >For instance Crafty & Junior (authors) are more interested in human competitions >than in computer competitions. And its not the same. > >cheers, >pavs I see no reason, then, why a chess engine programmer could not re-optimize his program for each new competition, assumming conditions changed enough to make a difference. [and, incidentally, optimize book, GUI maybe, etc.] If "everybody does it," then why not? Keeps a level playing field. Bob D.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.