CCRL 40/15 Testing Conditions (previously 40/40)

[Ray]

Have you tried to run Amoeba 3.1 with 4 cores using ChessGUI, which is what I use for all of my testing?
You set threads to 4, but it still only uses 1.

I tried it here with ChessGUI - and yes only uses 1 thread with amoeba-win64.exe

[Ray]

It seems to be a ChessGUI bug though. ChessGUI sends the threds=1 command to the engine, and this is why:

The config file says this:

[Amoeba 3.1 64-bit 4CPU]
Affinity=0:0
Hash=512
Threads=1
MultiPV=1
Log=false
UCI_AnalyseMode=false

Even though the UCI paremeters popup is this:

[Ray]

However I edited the engine parameters again and saved them down, that config file has corrected itself, and now it is using 4 threads. You could also manually edit that file (but best not to call up the UCI parameters via the GUI after that).

On my Octal: (50% is 4 of 8 threads)
.
.

[dejan]

Have you tried to run Amoeba 3.1 with 4 cores using ChessGUI, which is what I use for all of my testing?
You set threads to 4, but it still only uses 1.

ChessGUI is a Windows application. I would have to either try to install it on my Linux workstation using Wine, and also use Windows build of Amoeba... - Too much hassle. Seems like Ray found a workaround!

[Ray]

Yes, close the GUI, re-start it and edit the engine and re-save the parameters seems to work. If not, manually edit the GUI engine config file.

[dejan]

Graham, did you try what Ray suggests?

[Graham Banks]

Graham, did you try what Ray suggests?

Not yet, but I will.

[enginechessfan]

Where do I find the SF 10 benchmark you guys use to get an idea of what time control my PC would use?

[Ray]

Where do I find the SF 10 benchmark you guys use to get an idea of what time control my PC would use?

Form here:

https://www.dropbox.com/sh/75gzfgu7qo94 ... Br4r_RwhGa

[enginechessfan]

Thanks! Is there a certain position that I analyze for the benchmark?

[Graham Banks]

Thanks! Is there a certain position that I analyze for the benchmark?

Starting position.

[Pi4chess]

Hello, l am new there

I would like to bench my little machine with ccrl standard but i don't know how to. Hope you can help me.

My machine is a Rasperry Pi 4 with Linux 64 bits OS.
Crafty bench is for Windows users. I read that is also a SF 10 bench ? but i did not manage to find the howto.

I have already SF 10 in my little machine and can force with Arena Gui to use only 1 core if necessary.

Thx

[Ray]

In Windows you just double-click the Stockfish 10 exe and type "bench", Adapted time control for 40/15 would be/:


T minutes for 40 moves where T = Your time / 2054 x 15 (for 40/15)

[Pi4chess]

Thanks ! The internal bench of SF I 1 core or multi-core?

[Pi4chess]

Thanks ! The internal bench of SF I 1 core or multi-core?

Ok I managed to bench 1 core = 5 sec and 4 cores = 2,5 sec.
Thx !

[Ray]

Bench on 1 core. Just type bench, it will use the defaults. It is SF10 you should be using for that.

[xmchess]

Hi
it is written
Our time control is equivalent to 40 moves in 15 minutes on an Intel i7 4770k at 3.5GHz.
We use Stockfish 10 as a benchmark to determine the equivalent time control for particular machine.
I would like to test my engine and i have a question :
how is the benchmark measured ?
What is the score with this benchmark on an Intel i7 4770k at 3.5GHz.
thx
excuse me my english is not very good

[Gabor Szots]

Scroll upwards and you'll find the answer.

[Aggelos]

With which version of Stockfish 10 do you run the bench?
x32 - x64 - or modern?

[Graham Banks]

With which version of Stockfish 10 do you run the bench?
x32 - x64 - or modern?

Modern 64-bit for newer computers or the standard 64-bit exe for older computers.

[dkappe]

Is there a discussion somewhere about calibrating for lc0/gpu?

[Graham Banks]

Is there a discussion somewhere about calibrating for lc0/gpu?

I'll leave that for either Ray or Sergio to answer.

[Ray]

Is there a discussion somewhere about calibrating for lc0/gpu?

We don't currently calibrate for the GPU. Time control for games is simply set based on the CPU speed. That works so far because no two testers have had the same GPU. It would be a problem if two testers had the same GPU in machines with very different CPUs. I think CEGT put GPUs in their machines that are a sensible match for the CPU, and slow down the GPU via Afterburner or something so that the leela ratio on their machines is not vastly different.

There has been a lot of discussion in the CCRL member sub-forums in the past so I suggest you look there rather than carry on the discussion here.

[ThatsMe]

Please excuse my ignorance, surely there is a very simple explanation for this that I don't know but I have been wondering for many weeks why there is no participation and no results from Stockfish 14 in the CCRL 40/15 list (as of today, just checked again http://ccrl.chessdom.com/ccrl/4040/), why is the list so lagging behind?

[Graham Banks]

Please excuse my ignorance, surely there is a very simple explanation for this that I don't know but I have been wondering for many weeks why there is no participation and no results from Stockfish 14 in the CCRL 40/15 list (as of today, just checked again http://ccrl.chessdom.com/ccrl/4040/), why is the list so lagging behind?

https://www.computerchess.org.uk/ccrl/4 ... ons_only=1