Stockfish performing bad on ryzen cpu.

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
mwyoung
Posts: 2339
Joined: Wed May 12, 2010 8:00 pm

Re: Stockfish performing bad on ryzen cpu.

Post by mwyoung » Mon Sep 14, 2020 11:04 pm

awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
Professing themselves to be wise, they became fools,
take on me. Foes 0.

awsmak
Posts: 21
Joined: Sat Sep 05, 2020 3:29 am
Full name: Akshay Chacko

Re: Stockfish performing bad on ryzen cpu.

Post by awsmak » Mon Sep 14, 2020 11:31 pm

mwyoung wrote:
Mon Sep 14, 2020 11:04 pm
awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
time control is set to 2 seconds per move on the smartphone and 3 seconds per move on laptop. ponder is set on on both of them. i have conducted atleast 200 games trying different time controls, hash sizes and thread counts. every match results in either a draw or the smartphone win.
i will be soon creating and posting a pgn file with 50 games.

Milos
Posts: 3832
Joined: Wed Nov 25, 2009 12:47 am

Re: Stockfish performing bad on ryzen cpu.

Post by Milos » Tue Sep 15, 2020 12:42 am

awsmak wrote:
Mon Sep 14, 2020 11:31 pm
mwyoung wrote:
Mon Sep 14, 2020 11:04 pm
awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
time control is set to 2 seconds per move on the smartphone and 3 seconds per move on laptop. ponder is set on on both of them. i have conducted atleast 200 games trying different time controls, hash sizes and thread counts. every match results in either a draw or the smartphone win.
i will be soon creating and posting a pgn file with 50 games.
You cannot be serious, you are using ponder, so while you are manually entering the moves from one platform to the other engines are actually thinking??? That can skew things a lot coz you are certainly taking more time to enter the moves than what is your time control.

Nay Lin Tun
Posts: 628
Joined: Mon Jan 16, 2012 5:34 am

Re: Stockfish performing bad on ryzen cpu.

Post by Nay Lin Tun » Tue Sep 15, 2020 12:53 am

OP dont even know how to set up engine tournment properly.

OP should do the first step properly then discussion about technical details.

awsmak
Posts: 21
Joined: Sat Sep 05, 2020 3:29 am
Full name: Akshay Chacko

Re: Stockfish performing bad on ryzen cpu.

Post by awsmak » Tue Sep 15, 2020 4:15 am

Milos wrote:
Tue Sep 15, 2020 12:42 am
awsmak wrote:
Mon Sep 14, 2020 11:31 pm
mwyoung wrote:
Mon Sep 14, 2020 11:04 pm
awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
time control is set to 2 seconds per move on the smartphone and 3 seconds per move on laptop. ponder is set on on both of them. i have conducted atleast 200 games trying different time controls, hash sizes and thread counts. every match results in either a draw or the smartphone win.
i will be soon creating and posting a pgn file with 50 games.
You cannot be serious, you are using ponder, so while you are manually entering the moves from one platform to the other engines are actually thinking??? That can skew things a lot coz you are certainly taking more time to enter the moves than what is your time control.
I have tried atleast 20-30 games with ponder off. results doesn't change. Even if ponder is on and i take only less than 1.5sec to copy the moves. That will affect both engines equally right? I just want to know why my laptop is not winning a single game. Help me brother.

Nay Lin Tun
Posts: 628
Joined: Mon Jan 16, 2012 5:34 am

Re: Stockfish performing bad on ryzen cpu.

Post by Nay Lin Tun » Tue Sep 15, 2020 6:20 am

awsmak wrote:
Tue Sep 15, 2020 4:15 am
Milos wrote:
Tue Sep 15, 2020 12:42 am
awsmak wrote:
Mon Sep 14, 2020 11:31 pm
mwyoung wrote:
Mon Sep 14, 2020 11:04 pm
awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
time control is set to 2 seconds per move on the smartphone and 3 seconds per move on laptop. ponder is set on on both of them. i have conducted atleast 200 games trying different time controls, hash sizes and thread counts. every match results in either a draw or the smartphone win.
i will be soon creating and posting a pgn file with 50 games.
You cannot be serious, you are using ponder, so while you are manually entering the moves from one platform to the other engines are actually thinking??? That can skew things a lot coz you are certainly taking more time to enter the moves than what is your time control.
I have tried atleast 20-30 games with ponder off. results doesn't change. Even if ponder is on and i take only less than 1.5sec to copy the moves. That will affect both engines equally right? I just want to know why my laptop is not winning a single game. Help me brother.
No one can help you unless you
1.know yourself that you are horrible in technology
2. willing to learn and
3. correct yourself.

User avatar
yurikvelo
Posts: 570
Joined: Sat Dec 06, 2014 12:53 pm

Re: Stockfish performing bad on ryzen cpu.

Post by yurikvelo » Tue Sep 15, 2020 8:05 am

mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
There is no way Sf 12 should be getting only 1.5 mnps on 6 cores.
this is expected speed for NNUE @ Qualcomm Snapdragon 855
Level 3 seconds/move, ponder on
with Ponder=On and manual move enter - no one knows how much time actually each side get.
Basically both sides were running infinite analysis mode, waiting for human input and getting extra +3 seconds after random human input

corres
Posts: 3351
Joined: Wed Nov 18, 2015 10:41 am
Location: hungary

Re: Stockfish performing bad on ryzen cpu.

Post by corres » Tue Sep 15, 2020 8:43 am

yurikvelo wrote:
Tue Sep 15, 2020 8:05 am
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
There is no way Sf 12 should be getting only 1.5 mnps on 6 cores.
this is expected speed for Qualcomm CPU
AMD CPU-s are very sensitive things.
There are lot of factor what have effect on the power of an AMD CPU.
At first the power supply and its stability. It is especially important in the case of a mobile machine.
At second the cooling of the CPU. The power of an overheated CPU obviously smaller than a well cooled one.
At third setting in BIOS:
It is very comfortable using Ryzen Master to set the processor parameters, but it only works well, if the circumstances does not alter a lot. For e.g the room temperature. If you use Ryzen Master in a fresh room, it is possible your CPU will not work well in a dog-day.
There are such parameters in BIOS what you are obliged to set manually:
SMT(HT) ideally it enhances the power, but it also enhances the temperature of CPU, what decreases the power, so the result is ambiguous.
XMP-profile. In general it also enhance the full power the machine, bit enhance the temperature of CPU too.
The chosen optimization of Stockfish-binary (exe). AVX2 optimized exe give higher chess power, but it also enhance the temperature, so maybe better using popcount (modern) exe on a mobile machine.
It is also very important, when you run a chess engine, you should switch off every other application!
It is stupid things to watch film or use internet simultaneously with running a chess engine!
The most important parameter of Stockfish is the Hash. It needs at least 2Gb.
So for the full machine (64bits) it needs at least 8 Gb RAM.
Note
I use for playing chess only table PC in a well cooled case and I run only the chess engine alone!.

awsmak
Posts: 21
Joined: Sat Sep 05, 2020 3:29 am
Full name: Akshay Chacko

Re: Stockfish performing bad on ryzen cpu.

Post by awsmak » Tue Sep 15, 2020 11:02 am

Nay Lin Tun wrote:
Tue Sep 15, 2020 6:20 am
awsmak wrote:
Tue Sep 15, 2020 4:15 am
Milos wrote:
Tue Sep 15, 2020 12:42 am
awsmak wrote:
Mon Sep 14, 2020 11:31 pm
mwyoung wrote:
Mon Sep 14, 2020 11:04 pm
awsmak wrote:
Mon Sep 14, 2020 10:56 pm
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
awsmak wrote:
Mon Sep 14, 2020 6:33 am
Nay Lin Tun wrote:
Mon Sep 14, 2020 6:29 am
Why your Droidfish is running < 1.6 mnps, are you using 1 core? 4 cores droidfish should be around 5 mnps.
Stockfish 11 gives 5000-6000knps stockfish 12 gives only 1500knps using 6 cores
All I know is you have some kind of issue. The is no way Sf 12 should be getting only 1.5 mnps on 6 cores. I get almost 1 mnps on 1 core at the start position.

You are using a laptop, and it may not have good thermals. And AVX creates much more heat and stress on the CPU. Try running the other version of SF 12.
on my laptop on 8 cores sf12 gives 9+mnps. my smartphone on 6 cores sf12 gives 1.5-2mnps. but my laptop is always loosing or drawing against my smartphone.
That sounds more reasonalbe. 8 cores and 9 mnps. How many games have you played and what is the score? And 512mb hash is low amount to be using, but that should not affect the results much.

If it is at very short time control. I could see were their is not much difference. SF 12 is a sprinter and gains much Elo rapidly. But 8 cores / 9 mnps should be better then 6 cores / 2 mnps. But need to see how many games and score. Stats matter here.

And did you have ponder on one system, and ponder off on the other system. Just suggesting possibilities.
time control is set to 2 seconds per move on the smartphone and 3 seconds per move on laptop. ponder is set on on both of them. i have conducted atleast 200 games trying different time controls, hash sizes and thread counts. every match results in either a draw or the smartphone win.
i will be soon creating and posting a pgn file with 50 games.
You cannot be serious, you are using ponder, so while you are manually entering the moves from one platform to the other engines are actually thinking??? That can skew things a lot coz you are certainly taking more time to enter the moves than what is your time control.
I have tried atleast 20-30 games with ponder off. results doesn't change. Even if ponder is on and i take only less than 1.5sec to copy the moves. That will affect both engines equally right? I just want to know why my laptop is not winning a single game. Help me brother.
No one can help you unless you
1.know yourself that you are horrible in technology
2. willing to learn and
3. correct yourself.
I do brother. I ran the tournament by runing stockfish12 remotely from laptop using a server application and by using chess for android apk on smartphone which act as client. This way I made sure there is no input delay or anything. The tournament had a total of 10 games started from beginning. Level was 2 seconds per move. Out of 10 games stockfish on laptop won a single game and rest of the 9 games were draw. I conducted another tournament where both engines were ran in classic eval mode and all games resulted in draw.
In nnue mode
Laptop nps - 9mnps at start
Smartphone nps - 1.5mnps at start

In classic mode
Laptop nps -15mnps at start
Smartphone nps - 5mnps at start

Hash on laptop is set to 512MB
Hash on Smartphone is set to 128MB

awsmak
Posts: 21
Joined: Sat Sep 05, 2020 3:29 am
Full name: Akshay Chacko

Re: Stockfish performing bad on ryzen cpu.

Post by awsmak » Tue Sep 15, 2020 11:22 am

corres wrote:
Tue Sep 15, 2020 8:43 am
yurikvelo wrote:
Tue Sep 15, 2020 8:05 am
mwyoung wrote:
Mon Sep 14, 2020 10:41 pm
There is no way Sf 12 should be getting only 1.5 mnps on 6 cores.
this is expected speed for Qualcomm CPU
AMD CPU-s are very sensitive things.
There are lot of factor what have effect on the power of an AMD CPU.
At first the power supply and its stability. It is especially important in the case of a mobile machine.
At second the cooling of the CPU. The power of an overheated CPU obviously smaller than a well cooled one.
At third setting in BIOS:
It is very comfortable using Ryzen Master to set the processor parameters, but it only works well, if the circumstances does not alter a lot. For e.g the room temperature. If you use Ryzen Master in a fresh room, it is possible your CPU will not work well in a dog-day.
There are such parameters in BIOS what you are obliged to set manually:
SMT(HT) ideally it enhances the power, but it also enhances the temperature of CPU, what decreases the power, so the result is ambiguous.
XMP-profile. In general it also enhance the full power the machine, bit enhance the temperature of CPU too.
The chosen optimization of Stockfish-binary (exe). AVX2 optimized exe give higher chess power, but it also enhance the temperature, so maybe better using popcount (modern) exe on a mobile machine.
It is also very important, when you run a chess engine, you should switch off every other application!
It is stupid things to watch film or use internet simultaneously with running a chess engine!
The most important parameter of Stockfish is the Hash. It needs at least 2Gb.
So for the full machine (64bits) it needs at least 8 Gb RAM.
Note
I use for playing chess only table PC in a well cooled case and I run only the chess engine alone!.
Thanks brother. Mine is an Asus laptop. Ryzen master is not supported and there are no controls in bios to turn off multithreading(SMT). I always close the background apps when running a tournament or analysis mode. Hash is set to 512MB. There is no overheating problem or downclocking i have noticed it is an AC controlled room.

Post Reply