Act 3 Threadripper Results

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

Moderators: hgm, Rebel, chrisw

corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Act 3 Threadripper Results

Post by corres »

corres wrote: Fri Dec 27, 2019 11:40 pm
MikeB wrote: Fri Dec 27, 2019 9:50 am ...
engine: Black-Diamond-122719
...
...
You attached the source of Stockfish251119 as "Honey-XR7-src" in a zip file to Honey-XR7.
Where is the real source of Honey-XR7?
There some derivatives of Stockfish what are not welcomed on this site because their authors hurt the GPL and they do not come to light the source of their derivatives.
I think GPL is GPL so I am waiting for the source of Honey-XR7...
Paloma
Posts: 1167
Joined: Thu Dec 25, 2008 9:07 pm
Full name: Herbert L

Re: Act 3 Threadripper Results

Post by Paloma »

MikeB wrote: Sun Dec 29, 2019 4:40 am Geekbench results after some modest BIOS tweaks
https://browser.geekbench.com/v4/cpu/15082989

Currently ranked #100 on the all time 'Geekbench 4' listing:
https://browser.geekbench.com/v4/cpu/multicore?page=4

Some nice (and expensive) hardware are listed just before and after my listing.
User: DieHardFan
How is it possible that an Intel Xeon W-3175X 4700 MHz (28 cores) only 6337
on single core reached, despite 1GHz higher clock rate (4700 vs 3700)

But multi-core with 28 cores almost matches your value with 32 cores? (116571 vs 117738)

see item Jun 15, 2019
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Act 3 Threadripper Results

Post by MikeB »

corres wrote: Sun Dec 29, 2019 9:28 am
corres wrote: Fri Dec 27, 2019 11:40 pm
MikeB wrote: Fri Dec 27, 2019 9:50 am ...
engine: Black-Diamond-122719
...
...
You attached the source of Stockfish251119 as "Honey-XR7-src" in a zip file to Honey-XR7.
Where is the real source of Honey-XR7?
There some derivatives of Stockfish what are not welcomed on this site because their authors hurt the GPL and they do not come to light the source of their derivatives.
I think GPL is GPL so I am waiting for the source of Honey-XR7...
Sorry about that - it defaults to master branch and my bad , forgot to update it. Branch honey has honey source.
Image
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Act 3 Threadripper Results

Post by MikeB »

Paloma wrote: Sun Dec 29, 2019 11:26 am
MikeB wrote: Sun Dec 29, 2019 4:40 am Geekbench results after some modest BIOS tweaks
https://browser.geekbench.com/v4/cpu/15082989

Currently ranked #100 on the all time 'Geekbench 4' listing:
https://browser.geekbench.com/v4/cpu/multicore?page=4

Some nice (and expensive) hardware are listed just before and after my listing.
User: DieHardFan
How is it possible that an Intel Xeon W-3175X 4700 MHz (28 cores) only 6337
on single core reached, despite 1GHz higher clock rate (4700 vs 3700)

But multi-core with 28 cores almost matches your value with 32 cores? (116571 vs 117738)

see item Jun 15, 2019
I have no way of answering your question, but it is a very good question

Edit: Upon further reflection , I’m running with some modest oc settings - settings that are applied easily in the Bios that Asus suggests. So in single core mode , there would be no throttling on the threadripper , when in mutli core mode - you would get some throttling as you approach 85C. The Asus motherboard will not let temps go over 85 C. The max oc frequency for the 3970x is 4.5 GHz. My base frequency with the bios changes is about 37.25 or so. It’s on auto tune OC mode so it’s constantly changing
Image
Paloma
Posts: 1167
Joined: Thu Dec 25, 2008 9:07 pm
Full name: Herbert L

Re: Act 3 Threadripper Results

Post by Paloma »

Thanks for the explanation.
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Act 3 Threadripper Results

Post by corres »

MikeB wrote: Sun Dec 29, 2019 6:36 pm
corres wrote: Sun Dec 29, 2019 9:28 am
corres wrote: Fri Dec 27, 2019 11:40 pm
MikeB wrote: Fri Dec 27, 2019 9:50 am ...
engine: Black-Diamond-122719
...
...
You attached the source of Stockfish251119 as "Honey-XR7-src" in a zip file to Honey-XR7.
Where is the real source of Honey-XR7?
There some derivatives of Stockfish what are not welcomed on this site because their authors hurt the GPL and they do not come to light the source of their derivatives.
I think GPL is GPL so I am waiting for the source of Honey-XR7...
Sorry about that - it defaults to master branch and my bad , forgot to update it. Branch honey has honey source.
And where is "Branch honey"?
On https://github.com/michaelb7/stockfish/releases/tag/xr7 is source of Stockfish25112019 still.
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Act 3 Threadripper Results

Post by zullil »

corres wrote: Mon Dec 30, 2019 8:54 am
And where is "Branch honey"?
On https://github.com/michaelb7/stockfish/releases/tag/xr7 is source of Stockfish25112019 still.
https://github.com/MichaelB7/Stockfish/tree/honey
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Act 3 Threadripper Results

Post by corres »

zullil wrote: Mon Dec 30, 2019 1:00 pm
corres wrote: Mon Dec 30, 2019 8:54 am And where is "Branch honey"?
On https://github.com/michaelb7/stockfish/releases/tag/xr7 is source of Stockfish25112019 still.
https://github.com/MichaelB7/Stockfish/tree/honey
Thanks.
At last this is the true source of Honey
I use MinGW so I was afraid it would be issues during compilation but there were nothing.
Congrat the developers!
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Act 3 Threadripper Results

Post by MikeB »

corres wrote: Mon Dec 30, 2019 7:07 pm
zullil wrote: Mon Dec 30, 2019 1:00 pm
corres wrote: Mon Dec 30, 2019 8:54 am And where is "Branch honey"?
On https://github.com/michaelb7/stockfish/releases/tag/xr7 is source of Stockfish25112019 still.
https://github.com/MichaelB7/Stockfish/tree/honey
Thanks.
At last this is the true source of Honey
I use MinGW so I was afraid it would be issues during compilation but there were nothing.
Congrat the developers!
If you have a Unix shell in Windows , just change the compiler and ARCH at the top and run “build.sh” from the command line. Alternatively, use your normal make commands for Stockfish and add HONEY=yes for Honey , or BLAU=yes for Bluefish or NOIR=yes for Blackdiamond or WEAKFISH=yes for Weakfish. The caps and no spaces are important. Your normal make command will build the functionally equivalent Stockfish but with all the bells and whistles ( 4 opening books , Play by FIDE Elo, adaptive play and variety play options and short cut terminal command , e.g, ‘s thread 4’ is the same as ‘setoption name threads value 4’ or ‘s hash 16384’ equals ‘setoption name hash value 16384’ - total bench nodes will be the same as Stockfish.
adding FORTRESS_DETECT=yes , will add Fortress detection code to any of the builds. Enjoy!
Image
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Act 3 Threadripper Results

Post by corres »

MikeB wrote: Mon Dec 30, 2019 11:49 pm
corres wrote: Mon Dec 30, 2019 7:07 pm
zullil wrote: Mon Dec 30, 2019 1:00 pm
corres wrote: Mon Dec 30, 2019 8:54 am And where is "Branch honey"?
On https://github.com/michaelb7/stockfish/releases/tag/xr7 is source of Stockfish25112019 still.
https://github.com/MichaelB7/Stockfish/tree/honey
Thanks.
At last this is the true source of Honey
I use MinGW so I was afraid it would be issues during compilation but there were nothing.
Congrat the developers!
If you have a Unix shell in Windows , just change the compiler and ARCH at the top and run “build.sh” from the command line. Alternatively, use your normal make commands for Stockfish and add HONEY=yes for Honey , or BLAU=yes for Bluefish or NOIR=yes for Blackdiamond or WEAKFISH=yes for Weakfish. The caps and no spaces are important. Your normal make command will build the functionally equivalent Stockfish but with all the bells and whistles ( 4 opening books , Play by FIDE Elo, adaptive play and variety play options and short cut terminal command , e.g, ‘s thread 4’ is the same as ‘setoption name threads value 4’ or ‘s hash 16384’ equals ‘setoption name hash value 16384’ - total bench nodes will be the same as Stockfish.
adding FORTRESS_DETECT=yes , will add Fortress detection code to any of the builds. Enjoy!
Thanks for advice.
I used the base command of Stockfish after I delete the appropriate ### in Make file.