The AMD Threadripper 2990WX 32-Core and 2950X 16-Core Review
by Dr. Ian Cutress on August 13, 2018 9:00 AM ESTOverclocking: 4.0 GHz for 500W
Who said that a 250W processor should not be overclocked? AMD prides itself as being a processor manufacturer that offers every consumer processor as a multiplier unlocked part, as well as using a soldered thermal interface material to assist with thermal dissipation performance. This 2990WX has an X in the same, so let the overclocking begin!
Actually, confession time. We did not have much time to do overclocking by any stretch. This processor has a 3.0 GHz base frequency and a 4.2 GHz turbo frequency, and in an air-conditioned room using the 500W Enermax Liqtech cooler, when running all cores under POV-Ray, we observed each core running around 3150 MHz, which is barely above the turbo frequency. The first thing I did was set the all-core turbo to 4.2 GHz, the same as the single core turbo frequency. That was a bust.
However, the next stage of my overclocking escapades surprised me. I set the CPU to a 40x multiplier in the BIOS, for 4.0 GHz on all the cores, all the time. I did not adjust the voltage, it was kept at auto, and I was leaving the ASUS motherboard to figure it out. Lo and behold, it performed flawlessly through our testing suite at 4.0 GHz. I was shocked.
All I did for this overclock was turn a setting from ‘auto’ to ‘40’, and it breezed through almost every test I threw at it. I say almost every test – our Prime95 power testing failed. But our POV-Ray power testing, which draws more power, worked. Every benchmark in the suite worked. Thermals were high (in the 70s), but the cooler could take it, and with good reason too.
At full load in our POV-Ray test, the processor was listed as consuming 500W. The cooler is rated for 500W. At one point we saw 511W. This was split between 440W for the cores (or 13.8W per core) and 63W for the non-core (IF, IO, IMC) which equates to only 12.5% of the full power consumption. It answers the question from our Infinity Fabric power page - if you want the interconnect to be less of the overall power draw, overclock!
We also tried 4.1 GHz, and that seemed to work as well, although we did not get a full benchmark run out of it before having to pack the system up. As stated above, 4.2 GHz was a no-go, even when increasing the voltage. With tweaking (and the right cooling), it could be possible. For anyone wanting to push here, chilled water might be the way to go.
Performance at 4.0 GHz
So if the all-core frequency was 3125 MHz, an overclock to 4000 MHz all-core should give a 28% performance increase, right? Here are some of the key tests from our suite.
Overclocking the 2990WX is a mixed bag, because of how it does really well in some tests, and how it still sits behind the 2950X in others due to the bi-modal nature of the cores. In the tests were it already wins, it pushes out a lot more: Blender is up 19% in throughput, POV-Ray is up 19%, 3DPM is up 19%. The other tests, is catches back up to the 2950X (Photoscan), or still lags behind (app loading, WinRAR).
Overclocking is not the cure-all for the performance issues on the 2990WX, but it certainly does help.
Facebook
Twitter
RSS
171 Comments
View All Comments
just4U - Monday, August 13, 2018 - link
Ian, were you testing this with the CM Wraith Cooler? If not is it something you plan to review?Ian Cutress - Monday, August 13, 2018 - link
Most of the testing data is with the Liqtech 240 liquid cooler, rated at 500W. I do have data taken with the Wraith Ripper, and I'll be putting some of that data out when this is wrapped up.IGTrading - Monday, August 13, 2018 - link
To be honest, with the top of the line 32core model, it is interesting to identify as many positive effect cases as possible, to see if that entire set of applications that truly benefit of the added cores will persuade power users to purchase it.Like you've said, it is a niche of a niche and seeing it be X% faster of Y% slower is not as interesting as seeing what it can actually do when it is used efficiently and if this this makes a compelling argument for power users.
PixyMisa - Tuesday, August 14, 2018 - link
Phoronix found that a few tests ran much faster on Linux - for 7zip compression in particular, 140% faster (as in, 2.4x). Some of these benchmarks could improve a lot with some tweaking to the Windows scheduler.phoenix_rizzen - Wednesday, August 15, 2018 - link
It'd be interesting to redo these tests on a monthly basis after Windows/BIOS updates are done, to see how performance changes over time as the Windows side of things is tweaked to support the new NUMA setup for TR2.At the very least, a follow-up benchmark run in 6 months would be nice.
Kevin G - Monday, August 13, 2018 - link
Chiplets!The power consumption figures are interesting but TR does have to manage one thing that the high end desktop chips from Intel don't: off-die traffic. The amount of power to move data off die is significantly higher than moving it around on-die. Even in that context, TR's energy consumption for just the fabric seems high. When only threads are loaded, they should only be with dies with the memory controllers leaving two dies idle. It doesn't appear that the fabric is powering down while those remote dies are also powering down. Any means of watching cores enter/exit sleep states in real time?
I'd also be fun to see with Windows Server what happens when all the cores on a die are unplugged from the system. Consdiering the AMD puts the home agent on the memory controller on each die, even without cores or memory attached, chances are that the home agent is still alive consuming power. It'd be interesting to see what happens on Sky Lake-SP as well if the home agents on the grid eventually power themselves down when there is nothing directly connected to them. It'd be worth comparing to the power consumption when a core is disabled in BIOS/EFI.
I also feel that this would be a good introduction for what is coming down the road with server chips and may reach the high end consumer products: chiplets. This would permit the removal of the off-die Infinity Links for something that is effectively on-die throughout the cluster of dies. That alone will save AMD several watts. The other thing about chiplets is that it would greatly simplify Thread Ripper: only two memory controller chiplets would be to be in the package vs. four as we have now. That should save AMD lots of power. (And for those reading this comment, yes, Intel has chiplet plans as well.). The other thing AMD could do is address how their cache coherency protocols work. AMD has hinted at some caching changes for Zen 2 but lacks specificity.
gagegfg - Monday, August 13, 2018 - link
do not seem to exist more than once the 16 additional core of the 2990wx compared to the 2950xIan Cutress - Monday, August 13, 2018 - link
https://www.anandtech.com/bench/product/2133?vs=21...Chaitanya - Monday, August 13, 2018 - link
Built for scientific workload.woozle341 - Monday, August 13, 2018 - link
Do you think the lack of AVX512 is an issue? I might build a workstation soon for data processing with R and Python for some Fortran models and post-processing. Skylake-X looks pretty good wit its quad memory channels despite its high price.