Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Rybka Discussion / New:Rybka 4 Benchmark List !
1 2 3 4 Previous Next  
Poll Which is better hardware measuring method for chess ? (Closed)
Best (highest) kN/s 26 59%
Fixed Depth 18 41%
- - By Sedat Canbaz (****) [tr] Date 2010-10-31 00:57 Edited 2010-10-31 01:01
Hello Chess Friends !

I have started a newer Rybka 4 Benchmark List !!

Some Notes:
------------
•The Kn/s values are based exactly on same position played by Deep Rybka 4 engine
•Only the highest Kn/s per same processor have been taken
•Rybka 4 Benchmarks are acceptable !!

And i hope too see many hardwares-thanks in advance !

For Rybka 4 Benchmark:
http://sedatchess.110mb.com/index.php?p=1_72

Best Wishes,
Sedat Canbaz
Parent - - By mocha1961 (***) [us] Date 2010-10-31 04:29
can i use fritz11? and how to disable pondering in fritz11?
Parent - - By Sedat Canbaz (****) [tr] Date 2010-10-31 11:55

> can i use fritz11? and how to disable pondering in fritz11?


Thanks for your interest

Yes,sure...you need to click with right-mouse over engine and then deactivate the box

Parent - - By Sedat Canbaz (****) [tr] Date 2010-10-31 12:25
A note for Ruben Comes

Dear Ruben,

I plan to add your results:1331 kns


Can you confirm that you get same results with 128 Hashtable size and 10 Min game?

Best,
Sedat
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-31 21:19
It was 2048 hash and in same position used by Lukas.
I have same results in blitz games or more, but never use 128 Hashtable.
Parent - By ernest (****) [fr] Date 2010-11-01 16:32

> It was 2048 hash and in same position used by Lukas


Which one was that?
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-01 17:51
Hello Chess Friends !

Last Updates:
-------------
Intel Xeon X5650
Intel Core i7 980X
AMD Phenom II X6 1090T

For Rybka 4 BenchMark List:
http://sedatchess.110mb.com/index.php?p=1_72

Best Regards,
Sedat Canbaz
Parent - By Geomusic (*****) Date 2010-11-01 19:23
Lukas should rename that 200 core The LHC, for how much energy it must be using, are you up to 50 kilowatts/hr yet?
Parent - - By mindbreaker (****) [us] Date 2010-11-01 21:31
I suppose it is a little late, but I think you should have used a position where it makes the same move regardless.  My computer chooses at least 3 different moves.  This makes it take different amounts of time depending on which one it found.  Just a strait amount of time on ponder makes more sense...say 5 minutes, then just submission of the screenshot with 5 min showing.
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-01 22:36 Edited 2010-11-01 22:38

>I suppose it is a little late,  My computer chooses at least 3 different moves.


Yes...now its too late,unfortunately i can not change the conditions,sorry

Actually its true that Rybka plays different moves,but as far as i noticed it does not hurt the kns values
Note that we should not concentrate which move will be...

And what is important here,we need to check the kns values,not which move will be played...

If you really want to test it in slow time control,its ok,but then i suggest you to disable the tableabase
Otherwise,in slower time controls,Rybka 4 is starting to use the tablebeases and that hurts the Kn/s performance

For better results,you need to run the benchmark minimum 3-4 times,then you have more chances to get higher Kn/s values

Good luck,
Sedat
Parent - By jpqy (**) [be] Date 2010-11-02 07:28
"For better results,you need to run the benchmark minimum 3-4 times,then you have more chances to get higher Kn/s values"

Then you use the Hash memory data..so indeed next run will be faster or get higher nodes.

Each test should start by a reboot/restart from the program,then you get a better averange!

JP.
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-02 13:53
Hello again Dear Friends!

First of all,i'd like to say 'BIG Thanks' to:
Kburcham
Martin Thoresen
Mindbreaker

Last Updated Hardwares:
---------------------------
Intel 2x Xeon X5667 @4.5GHz
Intel Core i7 980X @4.41GHz
Intel Core Q6600   @3.45GHz

Note:Rybka 4 BenchMarks are Acceptable !

For BenchMark List/Details:
http://sedatchess.110mb.com/index.php?p=1_72

Best Regards,
Sedat Canbaz
Parent - - By ernest (****) [fr] Date 2010-11-02 13:58

> http://sedatchess.110mb.com/index.php?p=1_72


Poll

How many people have problems with that link?
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-02 18:35

> How many people have problems with that link?


I have no idea why you are unable to load any pages...i have no problems of visiting all links
Parent - By Akbarfan (***) [de] Date 2010-11-02 18:37
No problem here.
Parent - - By ernest (****) [fr] Date 2010-11-02 19:21
Now works!... :cool: no idea why...
So I saved the page!  :twisted:
Thanks, Sedat!
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-02 19:30
Ernest,i hope the good news will be go for long time :smile:

Best,
Sedat
Parent - - By Loboestepario (****) [us] Date 2010-11-02 18:35
Some of those reported speeds are quite remarkable. I would be very interested in reading their overclocking parameters/cooling methods. I wonder if many more would be interested in this information as well?
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-02 18:53 Edited 2010-11-02 18:59

> Some of those reported speeds are quite remarkable. I would be very interested in reading their overclocking parameters/cooling methods. I wonder if many more would be interested in this information as well?


Kburcham's pc info:
SR-2 motherboard
12 gigs ram
2x 5667 6 core CPU @ 4.5gig Westmere
2x 470 EVGA GPU
12 threads @4.5
Hyperthreading off
air cooled 2x Noctua 14's
BFG 1200 ex
2x Crucial 300 SSD in Raid O

Ruben's hardware:
http://www.computerschess.blogspot.com/

Martin's hardware:
http://www.tcec-chess.org/info.php

Best,
Sedat
Parent - By NATIONAL12 (Gold) [gb] Date 2010-11-04 00:29
Kburcham computer is not stable 24/7 as he admitted on talkchess.only stable at max maybe of 4.2.i use 2x Xeon 5680 at 4.2 24/7 and would take my build over his any day.Tony(BigBen,Wizard built it) for me.
Parent - - By retep1 (***) [de] Date 2010-11-02 17:58
no large pages?
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-02 18:31

> no large pages?


Both are acceptable on/off

Best,
Sedat
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-02 18:45 Edited 2010-11-02 18:48
(652 hits)

But unfortunately only a few benchmarks so far from Rybka Chess Forum ?!

Strange indeed...i think here is the right place of posting this thread

I have no patience too see your hardware results !!!

Best Regards,
Sedat Canbaz
Parent - - By jpqy (**) [be] Date 2010-11-02 18:57
Sedat..i have also try some test before on chess forums..but no responds from these people? Always comments and no act!

I get on sites where nobody play chess more then 38.900 hits..i think now more then 1000 results they send me..that's the way it should go.

And i get the results i want to see without discussing if this test is good or not!
People are kind,they are interested to try something new and it is fun..

You do a great work..the rest don't care.

Kind regards,
Jean-Paul.
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-02 19:06
Hello Jean-Paul !

Thanks a lot for your understanding and kind words

Best,
Sedat
Parent - - By TheHug (Bronze) [us] Date 2010-11-02 19:12
I know you probably prefer faster benchmarks. But I wanted to help to your results

I only have a Intel Pentium M Processor 1.30Ghz- 19kns Rybka 4 :sad:

A pretty sad number I know. For me it has done its job, I've had this computer for a long time and it has suited me.

Jimmy
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-02 19:17

>I know you probably prefer faster benchmarks. But I wanted to help to your results


No...i am interested in all hardwares

So if you interested... i will be very glad too see you on the list
For more details how to run a Rybka bench :
http://sedatchess.110mb.com/index.php?p=1_72

Regards,
Sedat
Parent - - By Jesse Gersenson (***) [cz] Date 2010-11-08 21:14
TheHug, you can likely put a 1.7 ghz Pentium M in your machine.

I recently upgrade my Pentium M to a 2.0 chip and am getting 50kn/s.

search ebay for: pentium SL6N5
Parent - - By TheHug (Bronze) [us] Date 2010-11-09 04:45
Thanks you for your comments. I didn't realize you could upgrade a laptop. Personally its not the big of a deal. I do oh ok with my AN, I only used it for my OTB stuff. You really shouldn't use such low hardware for CC play. But it is for fun for me. I have a unique style with finding some stocking tactical blows sometimes. The engines don't always see all the tactical blows
Parent - By Jesse Gersenson (***) [cz] Date 2010-11-09 09:06
TheHug, for what it's worth, in case you did want to give it a try, i sent you a private message about how to upgrade your cpu.
Parent - - By Felix Kling (Gold) [de] Date 2010-11-02 19:23
Sedat, I don't understand the way how this benchmark is done - I would expect a position and simply letting Rybka calculate for some time, independent of the GUI being used and using infinite analysis. You make it quite complicated with the screenshots and so on.
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-02 19:42
Dear Felix,

>I would expect a position and simply letting Rybka calculate for some time, independent of the GUI being used and using infinite analysis.


Ok...i will update my site about using infinite analysis are accepted too

>You make it quite complicated with the screenshots and so on


Actually my work is too much complicated,too many efforts i spent...maybe its time to get a medal from RybkaChess :smile:
Btw,i think it will not be so hard and it will not take many time for hardware user to capture a screen shot

Best,
Sedat
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-02 20:03
Rybka 4 Bench's Conditions have been updated:
4.Open your preferred GUI:Fritz/Aquarium/Arena/Shredder/ChessGUI/ChessPartner/Winboard...
7.Use Time Control:10 Minutes/Game (Infinite analysis are accepted too)

Greetings,
Sedat
Parent - - By Uly (Gold) [mx] Date 2010-11-02 21:42
Why would one need to download a 'Benchmarking Position'? Can't you just post the FEN on that page so that it can be copied and pasted?
Parent - - By Werewolf (*****) [gb] Date 2010-11-02 22:18

> Why would one need to download a 'Benchmarking Position'? Can't you just post the FEN on that page so that it can be copied and pasted?


and what's wrong with the starting position, why do we need a 'special' position at all?
Parent - - By Uly (Gold) [mx] Date 2010-11-03 01:18
Good point, if Rybka recalculates her node-count, it's unknown how a position like that would "warp" them.

I recall Bobby C talking about a position that would make Rybka's node-count skyrocket exponentially, so that the time spent on the position would always give something higher, using the starting position would solve any potential problems, besides making it easier to send a benchmark.
Parent - By Sedat Canbaz (****) [tr] Date 2010-11-03 18:07

> I recall Bobby C talking about a position ...


Sounds very good to me,in this way we will have possibility to compare both lists and i have no patience to see a such project
Parent - - By NATIONAL12 (Gold) [gb] Date 2010-11-04 00:33
on another thread BigBen posted over 1000 kN/s on my computer at start position without large pages.
Parent - - By Werewolf (*****) [gb] Date 2010-11-04 08:50
How much hash were you using and how long was it running to get that output?

Can you give a screen shot of your PC running for 5 min with analysis?

I take it you're running 2 x 6 Core i7...?
Parent - - By Werewolf (*****) [gb] Date 2010-11-04 10:29
?
There's no print out there
Parent - - By NATIONAL12 (Gold) [gb] Date 2010-11-04 11:36
there are screenshots which BigBen did.
Parent - - By Werewolf (*****) [gb] Date 2010-11-04 18:25
ah ok missed that he had your computer.

I'm thinking of upgrading depending on how good Bulldozer is. I want 3x my current system which gets about 450kn/s on the start with Rybka 4.

Where are you based these days? I'm just north of Birmingham.
Parent - By NATIONAL12 (Gold) [gb] Date 2010-11-04 18:51
still Bristol.:smile:
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-03 18:03

> Why would one need to download a 'Benchmarking Position'? Can't you just post the FEN on that page so that it can be copied and pasted?


Good idea...my brain sometimes is not enough for all,need assistants :smile:and thanks for your suggestion,as you know-for some people pgn is easier for use,for some people FEN
But i agree with you...and in the next update i will include FEN position too

Best,
Sedat
Parent - - By Uly (Gold) [mx] Date 2010-11-03 21:27
I didn't mean you including a FEN with the download, I meant you post the FEN in your page as text, like this:

3r2k1/2q4n/rp2Rp1Q/8/3P4/p1P4P/5PP1/R5K1 w - -

That way people can do the bench without needing to download anything.

Why this position and not the opening one?
Parent - - By Sedat Canbaz (****) [tr] Date 2010-11-03 22:00 Edited 2010-11-03 22:02

> I didn't mean you including a FEN with the download, I meant you post the FEN in your page as text, like this:


Its ok...i understood you very well already in the first your posting

And in the next update,the bench FEN position will be appeared (without need downloading)

BTW,a little info about current benchmarking opening position:it is just 304 kb and i guarantee that its without any threats (viruses,sparms,warms...) :smile:

Best,
Sedat
Parent - - By Felix Kling (Gold) [de] Date 2010-11-03 22:32
the problem is more like that the current benchmark is made artificially complicated :) in principle you could only give FEN and ask for the engine output + kN/s after a minute as I gave it to you. That is much easier.
Parent - - By Banned for Life (Gold) Date 2010-11-03 22:43
Expected time to a certain depth would be more useful, and if the engine is deterministic, even gives reproducible results. kn/s from the opening position isn't good because the value is not independent of the search depth.

Time to a certain depth is also better because it is documented by most GUIs...
Parent - By Felix Kling (Gold) [de] Date 2010-11-03 22:58
I totally agree :)
Up Topic Rybka Support & Discussion / Rybka Discussion / New:Rybka 4 Benchmark List !
1 2 3 4 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill