Hammering The AMD Ryzen 7 1800X With An Intense, Threaded Workload

[phoronix]

Phoronix: Hammering The AMD Ryzen 7 1800X With An Intense, Threaded Workload

Today I got around to running a very heavy/demanding, very real-world workload on the AMD Ryzen 7 1800X that I've been meaning to test with this Zen CPU...

http://www.phoronix.com/scan.php?page=news_item&px=Ryzen-7-1800X-OPM

[tildearrow]

Typo:

But with more than
two threads is when the Core i7 5960X showed its stength.

[sykobee]

Could you test SMT scaling with 16 threads please?

I presume that the benchmark is loading up cores first, so the 8T run should not have touched SMT.

How much is the 5950X system?

[Michael]

Could you test SMT scaling with 16 threads please?

I presume that the benchmark is loading up cores first, so the 8T run should not have touched SMT.

How much is the 5950X system?

Yes I will have 16 threads added shortly, didn't mean to post the article yet, but yeah the 16 thread results should be there in ~20 minutes.

[defaultUser]

Nice test. One possible source of improvement in general is to pin the processes (Is it easy to change threads to processes on the graphics?) to the cores. Usually this option is controlled by the mpiexec/run parameters

[Michael]

Could you test SMT scaling with 16 threads please?

I presume that the benchmark is loading up cores first, so the 8T run should not have touched SMT.

How much is the 5950X system?

16 threads are now up.

[maffblaster]

Could you test SMT scaling with 16 threads please?

I presume that the benchmark is loading up cores first, so the 8T run should not have touched SMT.

How much is the 5950X system?

Presently $1,069.65 on NewEgg. Ryzen is a bargain at the price point.

[M@yeulC]

Nice test, thanks for sharing it.
There is a visible degradation when going from 8 to 16 threads. Probably some semaphore issues in the program itself. Was the CPU usage always 100%?
I wonder if we could attribute this to the "core complex" architecture and its added latency? For this, we would need to see 8 threads running on only one core complex, and 16 threads on both.
It could be related to some FPU vs ALU strengths, too, depending on what the benchmark tests.

[chuckula]

This goes to show the strength of Intel's overall core integration strategy at heavy-duty workloads that aren't just L1-cache centric microbenchmarks.

If an AMD part from 2014 had the same margin of victory that we see here over a higher-clocked Intel part with an equivalent core count that was just released this month, then not one person here would be calling the Intel part good, even if it was somewhat cheap (although a $500 chip on a platform that has seen major motherboard support issues isn't exactly "cheap" in any book).

Incidentally, even if the 5960X out of a new box is still expensive, if you are smart you can find very good open box deals.

[chuckula]

Hey Michael, thanks for the in-depth analysis.

Two questions:
1. Do you know whether or not this package uses AVX heavily?

2. Would it be possible to show some power-consumption numbers during one of the longer runs that uses all of the cores? [just the 5960X and 1800X would be fine]