LCZero update (2)

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

Moderators: hgm, Rebel, chrisw

Werewolf
Posts: 1795
Joined: Thu Sep 18, 2008 10:24 pm

Re: LCZero update

Post by Werewolf »

CMCanavessi wrote:
But that happens if you have a prehistoric version! Get a newer lczero.exe here: https://ci.appveyor.com/project/glinsco ... /artifacts
Does that mean this website is out of date?

https://github.com/glinscott/leela-ches ... ng-Started
Werewolf
Posts: 1795
Joined: Thu Sep 18, 2008 10:24 pm

Re: LCZero update

Post by Werewolf »

jkiliani wrote:
Werewolf wrote:Would you be able to estimate the elo gain per doubling the number of rollouts?

Is it anywhere near a normal engine where doubling the amount of time yields around 50 elo?
It is probably more, but so far no-one has calculated a scaling curve for Leela Chess. In theory, NN based engines scale better than Alpha-Beta engines.It would be a very good idea to test that though, would you be interested?
I'm interested in helping the project, yes. I was planning on donating GPU time. But I'm also running test matches against a dedicated unit as a benchmark of how strong LCZ is.
nabildanial
Posts: 126
Joined: Thu Jun 05, 2014 5:29 am
Location: Malaysia

Re: LCZero update

Post by nabildanial »

CMCanavessi wrote:
nabildanial wrote:
CMCanavessi wrote:-p 0 and --noponder are useless (ponder is not supported yet)

You could try lczero.exe -n -w latest --threads X

where X would be 2, 3, 4 or whatever gives the best GPU usage. Just test and see how you get best performance.
But without -p 0 --noponder, Leela just stops at 800 nodes every time. I don't want it to stop thinking at any nodes.
But that happens if you have a prehistoric version! Get a newer lczero.exe here: https://ci.appveyor.com/project/glinsco ... /artifacts
Done that but after giving it a "go infinite" command, it still stops at 801th node.
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: LCZero update

Post by carldaman »

jkiliani wrote:
Werewolf wrote:OK, I've followed all steps properly and the command line seems to work now, it shows how many GFLOPS I'm running etc.

After it has completed the command line procedure, how do you actually get the engine to play?

I loaded lczero as an Arena engine (was that the right thing to do?) but it doesn't respond when I play a move.

I feel so close...thanks for your help.
Try starting the engine from the command line. If that's successful, it will show an output like

Using 1 thread(s).
Generated 1924 moves
Detecting residual layers...v1...64 channels...6 blocks.

At this point, enter "go" to prompt it to play a move. If that works, the rest is a matter of configuring the command line options in Arena properly, I saw some answers to that in this thread (or the predecessor thread anyway)
I also followed all the instructions, but I'm getting an error: "lczero.exe has stopped working"

Code: Select all

Problem signature:
  Problem Event Name:	APPCRASH
  Application Name:	lczero.exe
  Application Version:	0.0.0.0
  Application Timestamp:	5aaff4fa
  Fault Module Name:	ucrtbase.DLL
  Fault Module Version:	10.0.10240.16390
  Fault Module Timestamp:	55a5b718
  Exception Code:	40000015
  Exception Offset:	0000000000065a5f
  OS Version:	6.1.7601.2.1.0.768.3
  Locale ID:	1033
  Additional Information 1:	4622
  Additional Information 2:	46221c5614ab12b1e5be5763271a1d1b
  Additional Information 3:	6c59
  Additional Information 4:	6c5907e5aa1019ba7de108efab35f048
Any idea what might be wrong? Same error is occurring from the command line and also in Arena.
Win 7 64-bit here.
Thanks,
CL

Edit: when run from the command line I get past

Code: Select all

Generated 1924 moves 
Detecting residual layers...v1...64 channels...6 blocks. 
Initializing OpenCL.
OpenCL: clGetPlatformIDs
Then the crash message pops up. There's no prompt to enter a go command after that.
User avatar
CMCanavessi
Posts: 1142
Joined: Thu Dec 28, 2017 4:06 pm
Location: Argentina

Re: LCZero update

Post by CMCanavessi »

New official version released:

https://github.com/glinscott/leela-ches ... s/tag/v0.4

Finally includes a windows build with all the dlls, and a working windows CPU-Only build as well
Follow my tournament and some Leela gauntlets live at http://twitch.tv/ccls
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: LCZero update

Post by Nay Lin Tun »

What is the estimated efficiency of CPU version, 20%, 30% or higher than that?
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: LCZero update

Post by carldaman »

CMCanavessi wrote:New official version released:

https://github.com/glinscott/leela-ches ... s/tag/v0.4

Finally includes a windows build with all the dlls, and a working windows CPU-Only build as well
Thanks a lot, this is actually working! :D
jpqy
Posts: 550
Joined: Thu Apr 24, 2008 9:31 am
Location: Belgium

Re: LCZero update

Post by jpqy »

Thanks!

Put the windows version into a folder..click on LCZero.. and i get just a flashing command prompt window ,open,close?

Still not working..

JP.
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: LCZero update

Post by Laskos »

CMCanavessi wrote:New official version released:

https://github.com/glinscott/leela-ches ... s/tag/v0.4

Finally includes a windows build with all the dlls, and a working windows CPU-Only build as well
I have a weak video card, but I didn't expect that:
http://www.talkchess.com/forum/viewtopi ... 45&start=5

CPU version is performing much better. Is LCZero using the GPU card properly?
jkiliani
Posts: 143
Joined: Wed Jan 17, 2018 1:26 pm

Re: LCZero update

Post by jkiliani »

If you have cutechess_cli or some other software that can conduct chess engine tournaments, could you run a competition between the CPU and GPU versions at equal playouts? There has been talk on Github that GPU version is producing large error due to OpenCL calculations sometime, it would be cool if you could provide data on that (and share on Github).

Your tournament so far is a good start, but just means that this GPU just doesn't work well for you. We'd need equal playout settings (e.g. 800) to make this meaningful in evaluating whether the GPU calculation errors hurt playing strength a lot.