Zen4 / AM5 specifically (power to output / Core Count for D.C.)

[cellarnoise]

Zen4 looks good for performance per power and heat if set to lower power. The lower the better!

For this first gen / next socket / AMD5 stuff, the Zen4 (before Zen4 / 3D) and hopefully a Zen4 7950x 3D, I'll run it in a much better power / watt to output ratio. Likely most of the time under 140 watts? Many more watts during challenges?

I still may buy a Zen3 or 64 core Epyc or 2 older generation in single or double?, but that will be determined on the power, performance, and price factor with RAM and Motherboards included...

Zen4 with DDR5 and current motherboards are not great on price per performance, but if we can burn them down for 3 to 5 years? NOT BAD?

 

[Markfw]

When I get my 2 7950x boxes up, I will try to run WCG if it gives me tasks, and at the 142 watt config. I will let you know what happens. Hopefully, I can get units to benchmark.

 

[pututu]

Original link posted in [H].

Well most of us doing DC are familiar with undervolting and underclocking for optimum computational power efficiency whether it is cpu or gpu.

4.8GHz all core with Vcore 985mV

 

[StefanR5R]

Here are generation-to-generation comparisons, in German (but the bar graphs tell you the gist of it):
https://www.computerbase.de/2022-09/amd-ryzen-7950x-7900x-7700x-7600x-test/
First page looks at performance at fixed clock.
Second page shows stock performance (and indicates a certain loss of power efficiency), then investigates performance at reduced power limits (which increases power efficiency).

If you click on "Bearbeiten" ( = edit) in the diagrams, you can change the selection of applications on which to base the ratings. The button "Einträge" ( = entries) folds out further tested processor configurations.

 

[StefanR5R]

Other posters in the CPU subforum have linked to it before, but anyway:
mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > Zen4's AVX512 Teardown
An analysis by the author of y-cruncher. One of the points to take home: Although theoretical peak throughput of Zen3/Zen4 AVX-256 and Zen4 AVX-512 is the same clock-for-clock, moving an application to AVX-512 on Zen4 can reduce bottlenecks of the CPU's frontend ( = utilize the execution units better), and also reduce energy spent in the CPU's frontend ( = spend respectively more of the overall power budget in the execution units and elsewhere).