Dann Corbit's second version of Crafty17.11 benchmarked !

[Johan Havegheer]

Geschrieben von: / Posted by: Johan Havegheer at 13 June 2000 22:27:06:
Hello Dann,
Just downloaded the second crafty 17.11 tweaked version.
Benchmark :
Crafty 17.11 : 286.482 nps
DC tweaked version : 340.429 nps
Wow my PIII-566 256KB On die really likes your version Dann.
Johan

[Volker Pittlik]

Geschrieben von: / Posted by: Volker Pittlik at 14 June 2000 00:50:45:
Als Antwort auf: / As an answer to: Dann Corbit's second version of Crafty17.11 benchmarked ! geschrieben von: / posted by: Johan Havegheer at 13 June 2000 22:27:06:
[...]
YAC (yet another comparision)
Dual-Celeron 2*433@520 MHz
--------------------------------------
Crafty v17.10_smp (Corbit)
White(1): bench
Running benchmark. . .
Total nodes: 69248065
Raw nodes per second: 532677
Total elapsed time: 130
SMP time-to-ply measurement: 4.923077
--------------------------------------
Crafty v17.11 (Hyatt)
White(1): bench
Running benchmark. . .
......
Total nodes: 52419454
Raw nodes per second: 288018
Total elapsed time: 182
SMP time-to-ply measurement: 3.516484
Crafty v17.11_smp (Hyatt)
White(1): bench
Running benchmark. . .
......
Total nodes: 57971020
Raw nodes per second: 513017
Total elapsed time: 113
SMP time-to-ply measurement: 5.663717
--------------------------------------
Crafty v17.11 (Corbit)
White(1): bench
Running benchmark. . .
......
Total nodes: 52419454
Raw nodes per second: 294491
Total elapsed time: 178
SMP time-to-ply measurement: 3.595506
Crafty v17.11_smp (Corbit)
White(1): bench
Running benchmark. . .
......
Total nodes: 58723695
Raw nodes per second: 538749
Total elapsed time: 109
SMP time-to-ply measurement: 5.871560
--------------------------------------
What does "SMP time-to-ply measurement" mean exactly?
Volker

[Aaron]

Geschrieben von: / Posted by: Aaron at 14 June 2000 14:01:13:
Als Antwort auf: / As an answer to: Re: Dann Corbit's second version of Crafty17.11 benchmarked ! geschrieben von: / posted by: Volker Pittlik at 14 June 2000 00:50:45:
I'm oh so confused..With so many different complied versions floating around..
In the past the choice was simple either take The version on Dann's page or Dr Hyatts. For me SMP versions are irrelevant..
Now on top of that, you can choose versions that allow running 6 pieces TB and those that don't (I don't run 6 pieces TB,so don't use that version?)..Plus, I believe, there were some inital errors in the source code, so even Dr Hyatt's site was undated with different executables 3 times now..
And now I read on the Crafty list, that those on Dr Hyatt's site now are not complied by Eugene Nalimov, who usually complies them..!!
I'm too lazy to run the benchmarks myself.."Just bench right?" SO, I'll take it on faith that Dann's 2nd complied version is faster..But I still can't resist asking does the 2nd version, allow use of 6 piece TBs? Plus it's for non-SMPs righ?

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 14 June 2000 15:00:01:
Als Antwort auf: / As an answer to: Re: Dann Corbit's second version of Crafty17.11 benchmarked ! geschrieben von: / posted by: Aaron at 14 June 2000 14:01:13:

I'm too lazy to run the benchmarks myself.."Just bench right?" SO, I'll take it on faith that Dann's 2nd complied version is faster..But I still can't resist asking does the 2nd version, allow use of 6 piece TBs? Plus it's for non-SMPs righ?

Just access the ftpsite of Dann Corbit and read the names of the executables. It's very simple. Otherwise try
http://www.in-trier.de/~quisinsky/news.htm
Best wishes...
Mogens

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 14 June 2000 19:50:54:
Als Antwort auf: / As an answer to: Re: Dann Corbit's second version of Crafty17.11 benchmarked ! geschrieben von: / posted by: Aaron at 14 June 2000 14:01:13:

I'm oh so confused..With so many different complied versions floating around..
In the past the choice was simple either take The version on Dann's page or Dr Hyatts. For me SMP versions are irrelevant..
Now on top of that, you can choose versions that allow running 6 pieces TB and those that don't (I don't run 6 pieces TB,so don't use that version?)..Plus, I believe, there were some inital errors in the source code, so even Dr Hyatt's site was undated with different executables 3 times now..
And now I read on the Crafty list, that those on Dr Hyatt's site now are not complied by Eugene Nalimov, who usually complies them..!!
I'm too lazy to run the benchmarks myself.."Just bench right?" SO, I'll take it on faith that Dann's 2nd complied version is faster..But I still can't resist asking does the 2nd version, allow use of 6 piece TBs? Plus it's for non-SMPs righ?

I only compile for 6 piece tablebase files now. There is no speed penalty, and it only makes the executable about 50K larger. So the versions you download from my ftp site can use any number of tablebase files.



My FTP site