25 comments

[ 3.0 ms ] story [ 120 ms ] thread
For some things, like compressing files, they are really good. But game performance still benefits from strong single core where Intel is the best still, unfortunately.
I think choice of GPU matters more to game performance these days than the CPU does.
Not for some games, where low settings are used on purpose. The CPU is the bottleneck in those cases. cough Overwatch cough.
But if your budget is limited, it probably makes sense to go with an AMD CPU if it enables splurging on the GPU.
I play Overwatch at max settings (on a 2K screen) and my FX8370E + RX580 works very fine with it.
I doubt that hardware at that resolution on max settings is hitting much more than 60fps though - a lot of people have 144hz displays.
Honestly the price difference is frankly incredible.

2990WX - $1799

Xeon Platinum 8180M - $16999 to $17999.

The 2990WX doesn't support registered memory, which currently limits it to 128GB of RAM.

Nobody seriously looking at a Xeon 8180M would consider the 2990WX as a suitable alternative. The closest competitor from AMD is the Epyc 7601.

Epyc 7610 - $4800

Xeon Platinum 8180M - $16999 to $17999

Still 275% more expensive...

A crown in a select of one benchmark, which is even questionable if it's optimised for all and every CPU benchmarked.

Great title, OP.

Anandtech:"The AMD Threadripper 2990WX 32-Core and 2950X 16-Core Review"

https://www.anandtech.com/print/13124/the-amd-threadripper-2...

HN: https://news.ycombinator.com/item?id=17749738 ( 322 points )

--

Phoronix:"A Look At The Windows 10 vs. Linux Performance On AMD Threadripper 2990WX"

https://www.phoronix.com/scan.php?page=article&item=2990wx-l...

HN: https://news.ycombinator.com/item?id=17756266 ( 234 points )

I wish tech review sites would do a molecular dynamics benchmark for processors like these. It's pretty easy to set up just a simple MD simulation, and there'd be tens of us interested in seeing the results.
The Phoronix test suite is open source. If patches were provided, I feel confident the result would be new benchmarks included in his articles.
Interesting. I, for some reason, was unaware that the benchmarks were user sourced. It appears to mainly be a system for bash scripts to run the program and then automatically format the results, is that correct? There's a benchmark for LAMMPS, but it'd be interesting to get one up for Gromacs or CHARMM.
(comment deleted)
There are only AMD and Intel products in that list. I understand they have effectively ruled out competition with comprehensive patent coverage.
Nobody prevents anyone from creating competing CPU. It just won't have x86 instruction set.
> Nobody prevents anyone from creating competing CPU.

I'm pretty sure that Intel, AMD, ARM, IBM and a host of others would disagree if you actually tried.

It looks like VIA[1] are actually planning on bringing out an AMD64[2] compatible CPU- though I honestly wonder if it'll get any traction at all (it could be interesting in the low power space assuming it can even vaguely compete on price).

[1] https://www.theinquirer.net/inquirer/news/3023729/via-techno...

[2] Yeah, I know we're supposed to call it x86-64 these days, but lets take a moment and remember some of the brilliance AMD can be capable of. And thank the gods we're not stuck with Itanium.

This is an absurd claim:

- No single threaded performance benchmarks

- No performance/watt benchmarks

- No tests under various real world loads

For the kinds of workloads you'd buy one of these for I don't think single threaded performance is really relevant. Your other two points are good though.
Have you heard of Amdahl’s law?
Threadripper should win an award for product-naming.