An Extensive Look At The AMD Naples vs. Rome Power Efficiency / Performance-Per-Watt

[phoronix]

Phoronix: An Extensive Look At The AMD Naples vs. Rome Power Efficiency / Performance-Per-Watt

Since the AMD EPYC 7002 "Rome" series launch in August we have continue to be captivated by the raw performance of AMD's Zen 2 server processors across many different workloads as covered now in countless articles. The performance-per-dollar / TCO is also extremely competitive against Intel's Xeon Scalable line-up, but how is the power efficiency of these 7nm EPYC processors? We waited to deliver those numbers until having a retail Rome board for carrying out those tests and now after that and then several weeks of benchmarking, here is an extensive exploration of the AMD EPYC 7002 series power efficiency as well as a look at the peak clock frequencies being achieved in various workloads to also provide some performance-per-clock metrics compared to Naples.

http://www.phoronix.com/vr.php?view=28588

[schmidtbag]

The differences between 7601 and 7502 are the most interesting, since they're the most apples-to-apples between the generations. That being said, a 29% increase in efficiency is insanely good for one generation.

[Jabberwocky]

That's a lovely selection of CPUs you have there. By the way why is the CPU fan blowing air into the PSU?

Looks like the motherboard has some awesome and terrible features. U.2 and being my favorite feature and top PCI-E x16 being so close to RAM slots the worse IMO. Great to see the smaller slots' right-side border removed!

As for the benchmarks, one might call it epic... {ಠʖಠ}

[duby229]

That's a lovely selection of CPUs you have there. By the way why is the CPU fan blowing air into the PSU?

Looks like the motherboard has some awesome and terrible features. U.2 and being my favorite feature and top PCI-E x16 being so close to RAM slots the worse IMO. Great to see the smaller slots' right-side border removed!

As for the benchmarks, one might call it epic... {ಠʖಠ}

Yeah, you're definitely right about the CPU fan, pretty dumb.

[Michael]

That's a lovely selection of CPUs you have there. By the way why is the CPU fan blowing air into the PSU?

Due to the TR4/SP3 socket design for the mounting screws and its layout on this motherboard and the design of the Noctua heatsink, not really configurable -- at least for the optimal direction of pushing out the back. Only alternative would be pushing the air in the opposite direction as now.

[M@GOid]

That's a lovely selection of CPUs you have there. By the way why is the CPU fan blowing air into the PSU?

Looks like the motherboard has some awesome and terrible features. U.2 and being my favorite feature and top PCI-E x16 being so close to RAM slots the worse IMO. Great to see the smaller slots' right-side border removed!

As for the benchmarks, one might call it epic... {ಠʖಠ}

Yeah, I noticed it too. A different chassis or cooler might fix that tough.

[microcode]

Wow, that Apache benchmark (requests per second per MHz) is very cool. It basically says that it only takes 114 thousand cycles to serve an HTTP request on Zen 2, or 28μs per request at 4GHz.

[eva2000]

Thanks Michael your article just came at the right time as I was discussing Intel and AMD power draws on another forum so posted a link to your article

[_Robert]

potential alternative 4u heatsink :

https://store.supermicro.com/4u-active-amd-epyc-snk-p0064ap4.html

[coder]

Thanks for all the tests, but...

Geometric Mean Of All Test Results

...why not separate GeoMeans for the performance results and the efficiency results?