TalkChess.com

Geekbench score for 2x AMD 7742 (128 Cores)

https://browser.geekbench.com/v4/cpu/14484543

the score 193,554 is also the new all time high multicore score on Geekbench ...

https://browser.geekbench.com/v4/cpu/multicore ( then sort by multicore)

MikeB wrote: ↑Mon Sep 02, 2019 5:40 am Geekbench score for 2x AMD 7742 (128 Cores)

https://browser.geekbench.com/v4/cpu/14484543

the score 193,554 is also the new all time high multicore score on Geekbench ...

https://browser.geekbench.com/v4/cpu/multicore ( then sort by multicore)

and just for kicks, a comparison to my 2010 Mac Pro , acquired used for about $2000 a few years ago

https://browser.geekbench.com/v4/cpu/co ... e=13431134

MikeB wrote: ↑Mon Sep 02, 2019 5:30 am

Dann Corbit wrote: ↑Fri Aug 30, 2019 2:46 am ...

And the bottom graphic here shows that the very expensive AMD 7742 smokes the AMD 7601 and utterly destroys the Intel 8280 in terms of nodes per dollar with SF and AF.

...

Indeed,

If you compare the dual Xeon Platinum 8280 system at 125M nps to the projected 106M nps of 3rd gen TR 32-core system one wonders why anyone would ever spend an extra $18,000 on CPU's for the Intel system.

Regards,

Zenmastur

Zenmastur wrote: ↑Mon Sep 02, 2019 6:02 am

MikeB wrote: ↑Mon Sep 02, 2019 5:30 am

Dann Corbit wrote: ↑Fri Aug 30, 2019 2:46 am ...

And the bottom graphic here shows that the very expensive AMD 7742 smokes the AMD 7601 and utterly destroys the Intel 8280 in terms of nodes per dollar with SF and AF.

...

Indeed,

If you compare the dual Xeon Platinum 8280 system at 125M nps to the projected 106M nps of 3rd gen TR 32-core system one wonders why anyone would ever spend an extra $18,000 on CPU's for the Intel system.

Regards,

Zenmastur

The same reason why some people drive Lamborghini and others drive Austin cars.

Joost Buijs wrote: ↑Mon Sep 02, 2019 7:20 am

Zenmastur wrote: ↑Mon Sep 02, 2019 6:02 am

MikeB wrote: ↑Mon Sep 02, 2019 5:30 am

Dann Corbit wrote: ↑Fri Aug 30, 2019 2:46 am ...

And the bottom graphic here shows that the very expensive AMD 7742 smokes the AMD 7601 and utterly destroys the Intel 8280 in terms of nodes per dollar with SF and AF.

...

Indeed,

If you compare the dual Xeon Platinum 8280 system at 125M nps to the projected 106M nps of 3rd gen TR 32-core system one wonders why anyone would ever spend an extra $18,000 on CPU's for the Intel system.

Regards,

Zenmastur

The same reason why some people drive Lamborghini and others drive Austin cars.

That's not the point. The point is how much faster is the fastest car than a cheap car. In this case the difference in speed from a cheap car to the fastest car has been significantly reduced because the cheap car is now much faster.

Regards,

Zenmastur

Zenmastur wrote: ↑Mon Sep 02, 2019 4:00 pm

Joost Buijs wrote: ↑Mon Sep 02, 2019 7:20 am

Zenmastur wrote: ↑Mon Sep 02, 2019 6:02 am

MikeB wrote: ↑Mon Sep 02, 2019 5:30 am

Dann Corbit wrote: ↑Fri Aug 30, 2019 2:46 am ...

And the bottom graphic here shows that the very expensive AMD 7742 smokes the AMD 7601 and utterly destroys the Intel 8280 in terms of nodes per dollar with SF and AF.

...

Indeed,

If you compare the dual Xeon Platinum 8280 system at 125M nps to the projected 106M nps of 3rd gen TR 32-core system one wonders why anyone would ever spend an extra $18,000 on CPU's for the Intel system.

Regards,

Zenmastur

The same reason why some people drive Lamborghini and others drive Austin cars.

That's not the point. The point is how much faster is the fastest car than a cheap car. In this case the difference in speed from a cheap car to the fastest car has been significantly reduced because the cheap car is now much faster.

Regards,

Zenmastur

I'm pretty sure the real point involves attracting women. Not sure that applies to Intel vs AMD.

I think previous Epycs were only 64 cores. I didn't know they were making a 128 core version now. What does it mean for computer chess?

Leo wrote: ↑Mon Sep 02, 2019 9:36 pm I think previous Epycs were only 64 cores. I didn't know they were making a 128 core version now. What does it mean for computer chess?

It means that programmers really should start thinking how to best make use of so many cores. Hopefully folks will come up with something better than "Lazy SMP".

zullil wrote: ↑Mon Sep 02, 2019 9:41 pm

Leo wrote: ↑Mon Sep 02, 2019 9:36 pm I think previous Epycs were only 64 cores. I didn't know they were making a 128 core version now. What does it mean for computer chess?

It means that programmers really should start thinking how to best make use of so many cores. Hopefully folks will come up with something better than "Lazy SMP".

MCTS with virtual-loss is clearly a powerful parallelization methodology, and there are many other MCTS-parallelization strategies available for further research / development. Various MCTS-parallelization strategies is likely to be the way forward for most programmers. "Deep" and expensive neural nets have a big advantage here: they search fewer nodes so they can therefore do so more efficiently.

AlphaBeta on the other hand, searches many millions of nodes. But its becoming more and more clear that they search inefficiently: with time-to-depth not scaling as well as nodes-per-second. I think the solution is to discover a more work-efficient methodology, which can search all those millions-of-nodes per second, but with improved time-to-depth timing.

I don't think there's ever such a "thing".
A tyre is round, and it can't get any rounder.