+100 elo breakthrough in new rewritten Lco engine ( Ceres)

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

Moderators: hgm, Rebel, chrisw

User avatar
AdminX
Posts: 6340
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by AdminX »

Nordlandia wrote: Sun Jan 03, 2021 8:24 pm Unfortunately Ceres only support 1-Core dedicated for the GPU at the moment.
Image
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
User avatar
AdminX
Posts: 6340
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by AdminX »

Has anyone else checked out it's real time monitoring tool.

Image
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
User avatar
cc2150dx
Posts: 325
Joined: Sat Nov 30, 2013 9:51 am
Location: Canada
Full name: Jason Coombs

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by cc2150dx »

mwyoung wrote: Mon Jan 04, 2021 2:16 am
We will fix that today. Never leave performance on the table. When it is free!

Memory should be easily OC to over 8000 Mhz give or take. GPU is good with around 100 Mhz increase. You will know if it crashes in your games. No harm will happen, and just lower the clock.

Memory OC is good and can be left on at all times when you find the OC stability. It will not crash Lc0 or Cares, and other games.

Just download MSI afterburner. This will work on all RTX cards.

Make sure you use the maximum power limit allowed, and temp limit allowed.

https://www.msi.com/Landing/afterburner
Thanks for the info :) I will look into it.
User avatar
pohl4711
Posts: 2435
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by pohl4711 »

After installing Cuda 11.2 by NVIDIA and the DotNet5-runtime environment by Microsoft, I succeeded running Ceres 0.80, using Chad's boxed version:


(This is not the slower debug-version! Using a 30x384 net, Ceres is around +18% faster, than Lc0 0.26.3 on my RTX 2060)

Now, I will do a NN-testrun with Ceres 0.80 66680 net (Lc0 0.26.3 has the highscore, using that net, in my NN-testings - will Ceres 0.80 be stronger?). Result in 3 days, if all works correctly...

https://www.sp-cc.de/nn-vs-sf-testing.htm
User avatar
jshriver
Posts: 1342
Joined: Wed Mar 08, 2006 9:41 pm
Location: Morgantown, WV, USA

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by jshriver »

Image
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by Nay Lin Tun »

Two things.
1. They haven't released windows version of Ceres yet.
2. Current Leela running on TCEC already implemented Ceres idea (2x speed on that hardwares, slower in slower hardware).You may consider established stable Leela instead.
bmp1974
Posts: 74
Joined: Wed Dec 04, 2019 11:25 am
Full name: Prasanna Bandihole

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by bmp1974 »

But the question is when Leela team is releasing the next version incorporating Ceres change. We are still stuck with v0.26.3.
Also there was a pull in github for using 2 nets one general and another smaller net exclusively for endgames. Its time they bring out the new release
Dann Corbit
Posts: 12540
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: +100 elo breakthrough in new rewritten Lco engine ( Ceres)

Post by Dann Corbit »

There is nothing wrong with the LC0 team waiting until they feel comfortable before making a release.
If we get our bun in a knot because we want it now, we can always pull the source and build it ourselves.

It is like the SF official releases. There are long, long waits between new OFFICIAL versions of SF.
But the code is there for any instantaneous change and there are even daily builds if you are not afraid of using them.

There are many valid reasons for waiting to reach certain goals before making an official release.
Suppose, for instance, that every daily SF release was an official release of SF.
Now, imagine that you are a tester. Can you keep up? Not a chance.

Now, when a software change is made, a responsible development team wants to be sure that the code is actually improved.
They can know this by careful testing.
Careful testing takes time.

They also want to know that they did not introduce something bad or dangerous.
They can know this by careful testing.
Careful testing takes time.

I bet you did not know that software was like love.

"Love takes time
Yes it's hard to find
You gotta take some time
To let love grow"
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.