SedatChess

[Homayoun]

Thanks , Mr. Sedat for your great works. Very interesting tests in both books and engines fields. you choose your subjects very intelligently.

[Sedat Canbaz]

Thanks , Mr. Sedat for your great works. Very interesting tests in both books and engines fields. you choose your subjects very intelligently.

Thank you Mr Homayoun

Yes...sometimes,
We have to do our best...otherwise
Some people (including myself) may have confusions,
Because, this is my 1st hash-table test on 2x Epyc...

But it is clear that,
1 core, 128 MB hash is not bad at all.. at least on this machine!

Greetings

[IbaiBuR]

Hello there )

1st of all,
This test is done due to some people have no clue
What is going on...and really sorry to say that,
For example, like Deeds member, for more details:
viewtopic.php?f=4&t=3618

SCCT Hash-Table Test: 128 MB Vs 512 MB

Elo difference: 6 (in favor for 128 MB)

Code: Select all

1 RapTora 2.2 128mb  +23/-16/=361 50.88%  203.5/400
2 RapTora 2.2 512mb  +16/-23/=361 49.13%  196.5/400

Games:
https://mega.nz/file/XlI2lAza#842Rgurpn ... vfD4Epfwv0

Draw-Ratio (based on 400 games) is not so high 90%

Conditions:
2x Epyc 7B12, CuteChess, Ponder OFF, 1 Core, Unique Book, 1m+1s, 4-MEN
Note that all matches are played via GUI's concurrent option: 64
And for each player is installed another folder copy of RaptoRa engine
I mean that both current players are used to play from different eng exe

Note also that:
These results will be published later on SedatChess site, because
Working on two sites is not so easy as it looks.. thanks for understanding

Greetings

This test is with Large Pages enabled or not?

[Amos 4ever]

Hello there )

1st of all,
This test is done due to some people have no clue
What is going on...and really sorry to say that,
For example, like Deeds member, for more details:
viewtopic.php?f=4&t=3618

SCCT Hash-Table Test: 128 MB Vs 512 MB

Elo difference: 6 (in favor for 128 MB)

Code: Select all

1 RapTora 2.2 128mb  +23/-16/=361 50.88%  203.5/400
2 RapTora 2.2 512mb  +16/-23/=361 49.13%  196.5/400

Games:
https://mega.nz/file/XlI2lAza#842Rgurpn ... vfD4Epfwv0

Draw-Ratio (based on 400 games) is not so high 90%

Conditions:
2x Epyc 7B12, CuteChess, Ponder OFF, 1 Core, Unique Book, 1m+1s, 4-MEN
Note that all matches are played via GUI's concurrent option: 64
And for each player is installed another folder copy of RaptoRa engine
I mean that both current players are used to play from different eng exe

Note also that:
These results will be published later on SedatChess site, because
Working on two sites is not so easy as it looks.. thanks for understanding

Greetings

This test is with Large Pages enabled or not?

LP is enabled by default in the code, I don't think the question makes sense, you simply either have user rights enabled in the OS or not

[deeds]

This test is with Large Pages enabled or not?

It is activated by default if available on brainlearn-like engines as raptora. Don't be confused by its renamed experience.bin file to exp file. As expected, this test has nothing to do with Khalid's advice on minimal hash size when a "sugar experience v2" file is used.

[IbaiBuR]

Hello there )

1st of all,
This test is done due to some people have no clue
What is going on...and really sorry to say that,
For example, like Deeds member, for more details:
viewtopic.php?f=4&t=3618

SCCT Hash-Table Test: 128 MB Vs 512 MB

Elo difference: 6 (in favor for 128 MB)

Code: Select all

1 RapTora 2.2 128mb  +23/-16/=361 50.88%  203.5/400
2 RapTora 2.2 512mb  +16/-23/=361 49.13%  196.5/400

Games:
https://mega.nz/file/XlI2lAza#842Rgurpn ... vfD4Epfwv0

Draw-Ratio (based on 400 games) is not so high 90%

Conditions:
2x Epyc 7B12, CuteChess, Ponder OFF, 1 Core, Unique Book, 1m+1s, 4-MEN
Note that all matches are played via GUI's concurrent option: 64
And for each player is installed another folder copy of RaptoRa engine
I mean that both current players are used to play from different eng exe

Note also that:
These results will be published later on SedatChess site, because
Working on two sites is not so easy as it looks.. thanks for understanding

Greetings

This test is with Large Pages enabled or not?

LP is enabled by default in the code, I don't think the question makes sense, you simply either have user rights enabled in the OS or not

Didnt even ask you, I know by far that it is enabled in the code, your reply doesnt make sense either as I am clearly asking if he has enabled the Lock Pages in the OS, thats all. Dont try to make me look as an stupid person / disqualify my question

[Amos 4ever]

This test is with Large Pages enabled or not?

LP is enabled by default in the code, I don't think the question makes sense, you simply either have user rights enabled in the OS or not

Didnt even ask you, I know by far that it is enabled in the code, your reply doesnt make sense either as I am clearly asking if he has enabled the Lock Pages in the OS, thats all. Dont try to make me look as an stupid person / disqualify my question

You're just specifying it now
no one called you stupid
but if you don't specify in your question that you mean in the OS, it makes nosense to the reader

[IbaiBuR]

If the reader already knows it is enabled by default in the code, then I think the user can also suspect that I am talking about the OS part, right? It is logical! But ok...

[Amos 4ever]

If the reader already knows it is enabled by default in the code, then I think the user can also suspect that I am talking about the OS part, right? It is logical! But ok...

In any case Lp does not affect the result but only increases the speed, but as you know with NNUE it has no influence
Previously gave a little help with engines without NNUE

[IbaiBuR]

If the reader already knows it is enabled by default in the code, then I think the user can also suspect that I am talking about the OS part, right? It is logical! But ok...

In any case Lp does not affect the result but only increases the speed, but as you know with NNUE it has no influence
Previously gave a little help with engines without NNUE

Yeah, I know, but for experience files for example it shold increase performance a bit too since allocating LargePage memory for the experience data will speed up I/O operations during training for example. Or am I wrong here?