Just ordered a Crucial C300 256GB ssd

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
M ANSARI
Posts: 3726
Joined: Thu Mar 16, 2006 7:10 pm

Re: Just ordered a Crucial C300 256GB ssd

Post by M ANSARI »

With EGTB's the probing most certainly hurts an engine's strength, but man oh man, how horrible it feels when an engine loses or fails to win because it is missing those EGTB's. I think SSD's alleviate the performance somewhat and that might change the attitude of many programmers to re-think how aggressive their EGTB probing code should be. I know Vas thinks that EGTB's bring no ELO gain and might even drop a few ELO's ... but that is based on thousands on fast games. I know of at least 2 wins that Rybka missed in major competition due to lack of EGTB's ... so when it does happen the supposedly lack of ELO gain rings hollow. Of course the best option is to do something like bitbases, where only the important data is probed and thus the probing code can be much more aggressive and on RAM it comes with practically no performance hit.
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: Just ordered a Crucial C300 256GB ssd

Post by michiguel »

M ANSARI wrote:With EGTB's the probing most certainly hurts an engine's strength,
I may understand that some people may think that EGTBs bring little or negligible increase in ELO. But decrease?

but man oh man, how horrible it feels when an engine loses or fails to win because it is missing those EGTB's. I think SSD's alleviate the performance somewhat and that might change the attitude of many programmers to re-think how aggressive their EGTB probing code should be. I know Vas thinks that EGTB's bring no ELO gain and might even drop a few ELO's ... but that is based on thousands on fast games.
at fixed depth, which is an unbelievably wrong way to test a pruning method like EGTBs!

I know of at least 2 wins that Rybka missed in major competition due to lack of EGTB's ... so when it does happen the supposedly lack of ELO gain rings hollow.
that is when the crude reality bites you. Imagine this is soccer. You lost two games in different World Cups in penalty kicks and your are the coach trying to explain that it is not worth practicing them because the probability of winning a game based on those is 0.5%.

Miguel

Of course the best option is to do something like bitbases, where only the important data is probed and thus the probing code can be much more aggressive and on RAM it comes with practically no performance hit.
User avatar
M ANSARI
Posts: 3726
Joined: Thu Mar 16, 2006 7:10 pm

Re: Just ordered a Crucial C300 256GB ssd

Post by M ANSARI »

Yes, the Brazilian team were notorious for having terrible penalty kickers except for the designated penalty kicker. But in the last several world cups a lot of emphasis has been on practicing many days simply kicking penalty kicks, and it has helped them win more World Cups.