Strategic Test Suite - STS - Deep Rybka 4

[Zherkovv]

Where can i find the results of "Deep Rybka 4" in STS ?

[Swaminathan]

Where can i find the results of "Deep Rybka 4" in STS ?

Hi, I don't have Deep Rybka 4. I hope those that have can run the test. I'd be interested in the result.

[Taner Altinsoy]

I can test rybka but in order for us to compare the results to others I have to run the test for other engines as well since our systems are quite different. I'll probably run the test for many of the leading engines on my laptop when I'm done with my tournament tomorrow. Then I guess it would take me a few days to run the tests.

Taner

[Swaminathan]

Thanks, I'm looking forward to it.

I think Jeremy had run such a test - You can find the post at programming forum.

[Taner Altinsoy]

I've run the test for Deep Rybka4 x64 and Houdini 1.02 x64 on my desktop which is AMD x2 4200 3GB RAM running Windows 7. SInce I've messed up with STS Stats I've added up results manually. Unless I didn't make a mistake results are below. Hopefully I'll manage STS stats better next time.

STS1 STS2 STS3 STS4 STS5 STS6 STS7 STS8 STS9 Total
Houdini 1.02 x64 94 85 91 84 87 82 86 76 77 87 849
Deep Rybka4 x64 92 91 86 87 90 88 82 68 77 76 837

regards,
Taner

[Taner Altinsoy]

I've run the test for Deep Rybka4 x64 and Houdini 1.02 x64 on my desktop which is AMD x2 4200 3GB RAM running Windows 7. SInce I've messed up with STS Stats I've added up results manually. Unless I didn't make a mistake results are below. Hopefully I'll manage STS stats better next time.

STS1 STS2 STS3 STS4 STS5 STS6 STS7 STS8 STS9 Total
Houdini 1.02 x64 94 85 91 84 87 82 86 76 77 87 849
Deep Rybka4 x64 92 91 86 87 90 88 82 68 77 76 837

regards,
Taner

I have completed the test for Houdini 1.01x64 as well. The results for all there is below.

STS1 STS2 STS3 STS4 STS5 STS6 STS7 STS8 STS9 STS10 Total
Houdini 1.01 x64 93 87 86 82 85 84 85 82 85 90 859
Houdini 1.02 x64 94 85 91 84 87 82 86 76 77 87 849
Deep Rybka4 x64 92 91 86 87 90 88 82 68 77 76 837

[noctiferus]

Hi, Swami,
as the other testers I tried the STS stats under Arena, following your instructions, but what I got is simply, as the others got, the engine hits, BD not the scores for non-hit moves, that is, IMHO, an important information.
Can you explain to a dummy, in more detail, how to get this info?
Thx
Enrico

[Swaminathan]

Hi, Swami,
as the other testers I tried the STS stats under Arena, following your instructions, but what I got is simply, as the others got, the engine hits, BD not the scores for non-hit moves, that is, IMHO, an important information.
Can you explain to a dummy, in more detail, how to get this info?
Thx
Enrico

Hi Enrico, I'm not I understand your message, can you explain it to me again? Thanks.

[Swaminathan]

I've run the test for Deep Rybka4 x64 and Houdini 1.02 x64 on my desktop which is AMD x2 4200 3GB RAM running Windows 7. SInce I've messed up with STS Stats I've added up results manually. Unless I didn't make a mistake results are below. Hopefully I'll manage STS stats better next time.

STS1 STS2 STS3 STS4 STS5 STS6 STS7 STS8 STS9 Total
Houdini 1.02 x64 94 85 91 84 87 82 86 76 77 87 849
Deep Rybka4 x64 92 91 86 87 90 88 82 68 77 76 837

regards,
Taner

I have completed the test for Houdini 1.01x64 as well. The results for all there is below.

STS1 STS2 STS3 STS4 STS5 STS6 STS7 STS8 STS9 STS10 Total
Houdini 1.01 x64 93 87 86 82 85 84 85 82 85 90 859
Houdini 1.02 x64 94 85 91 84 87 82 86 76 77 87 849
Deep Rybka4 x64 92 91 86 87 90 88 82 68 77 76 837

Hi Taner,

Thanks for the tests, I will test Houdini 1.01 and post results tomorrow.

This kind of thing happens. I believe it's something to do with the speed of the analysis. I notice that speed of Houdini 1.02 style of loading the analysis lines is different to 1.01. Perhaps that might have triggered some move changes in the last second. Also, are you sure that no other program was taking up memory while this was going on? Is this 1 CPU engine test?

[Taner Altinsoy]

Hi Swami,

I have run the tests on 2 core AMD 4200. I'm not sure but it has sth to do with arena GUI I think. I've run the test for the same engines and don't seem to be able to get the same results. I think 10 per position is too short for not to get affected by background processes etc. Also arena sometimes just skips the position at 3rd second as unsolved and adds 10 secs to the total time as if it has used that 10 secs for thinking. I'm trying to diagnose the problem but so far I have never been able to run the test for the same and get exactly same numbers. It's usually +/- 2 for each 100 positions, sometimes more. This is only happening in 10 sec mode. Will test with 20 or 30 secs per move. Also do you suggest using "activate abort analysis" in options tab? It definitely makes things much faster but you might have issues mentioned above like engine skippping without analysing, tho this needs confirmation whether it's the real cause.

Taner