The AMD Ryzen Threadripper 3960X and 3970X Review: 24 and 32 Cores on 7nm
by Dr. Ian Cutress, Andrei Frumusanu & Gavin Bonshor on November 25, 2019 9:05 AM ESTTest Bed and Setup
As per our processor testing policy, we take a premium category motherboard suitable for the socket, and equip the system with a suitable amount of memory running at the manufacturer's maximum supported frequency. This is also typically run at JEDEC subtimings where possible. It is noted that some users are not keen on this policy, stating that sometimes the maximum supported frequency is quite low, or faster memory is available at a similar price, or that the JEDEC speeds can be prohibitive for performance. While these comments make sense, ultimately very few users apply memory profiles (either XMP or other) as they require interaction with the BIOS, and most users will fall back on JEDEC supported speeds - this includes home users as well as industry who might want to shave off a cent or two from the cost or stay within the margins set by the manufacturer. Where possible, we will extend out testing to include faster memory modules either at the same time as the review or a later date.
Test Setup | |
AMD TR3 | Threadripper 3970X Threadripper 3960X |
Motherboard | ASUS ROG Zenith II Extreme (BIOS 0601) |
CPU Cooler | Thermaltake Riing 360 CLC |
DRAM | Corsair Dominator Platinum RGB 8x8 GB DDR4-3200 |
GPU | MSI GTX 1080 Gaming 8G |
PSU | Corsair AX860i |
SSD | Crucial MX500 2TB |
OS | Windows 10 1909 |
For our motherboards, we are using the latest firmware. It should be noted that our Intel tests do not have the latest Intel security updates for JCC and others, as the motherboard vendors for the models we used have not implemented them yet.
The latest AMD TR3 benchmarks were run by Gavin Bonshor, while I attended Supercomputing in Denver last week. Unfortunately both Intel and AMD decided to sample processors before the annual trade show conference, with launches only a couple of days after the show finished. As a result, our testing has been split between Gavin and myself, and we have endeavored to ensure parity through my automated testing suite.
Also, our compile test seems to have broken itself when we used Windows 10 1909, and due to travel we have not had time to debug why it is no longer working. We hope to get this test up and running in the new year, along with an updated test suite.
We must thank the following companies for kindly providing hardware for our multiple test beds. Some of this hardware is not in this test bed specifically, but is used in other testing.
245 Comments
View All Comments
schujj07 - Tuesday, November 26, 2019 - link
Where I work we now have 4x Dual 32 Core Epyc 7502s and 2x Dual 24 Core Epyc 7401s. We cannot move to Server 2016/2019 due to the per core licensing. However, for our VMware environment it is amazing how many VMs just 1 of those hosts can run.Supercell99 - Tuesday, November 26, 2019 - link
Is vmware stable on the new Epycs? I have some older Dells R630 2697 x2 I need to upgrade running ESXi 6.0 A bit nervous about jumping to AMD for production on vmware.schujj07 - Tuesday, November 26, 2019 - link
They are perfectly stable. We are running them for production work. 2nd Gen Epyc is only supported on 6.7 U3.Foeketijn - Tuesday, November 26, 2019 - link
On Epyc. Not TR. I would think.twtech - Monday, November 25, 2019 - link
Speaking of which, why does this review have so many gaming benchmarks, and say, no compiler benchmarks? I'd have liked to see the 32-core TR vs. the 3175x or 3275 compiling a large C++ project.eek2121 - Monday, November 25, 2019 - link
Not only that, but Anandtech is still doing gaming benchmarks on a Geforce 1080. Gamers Nexus has a much more production oriented review, but still no compiler benchmarks, etc.Slash3 - Tuesday, November 26, 2019 - link
I've never understood why AT has kept the GTX 1080. For purposes of benchmarking, it acts as an immediate bottleneck on faster CPUs and adds no value to a processor evaluation except in extreme cases such as the 2970WX/2990WX where performance impacts are made more readily evident. Even then, one or two simple tests would be enough to paint the picture, unless it called for further testing.It's simply a waste of benchmark time and continues to baffle me with its inclusion. The only reason I can think to keep it in reviews is to pad the Bench database, or that the tests can be completed quickly and it's simply spare time. I love AT, but sometimes they just make me scratch my head.
imaheadcase - Tuesday, November 26, 2019 - link
1080p is fine..they are using it for CPU benchmarks to bottleneck, not gpu.peevee - Tuesday, November 26, 2019 - link
It is GTX1080, not 1080p.DannyH246 - Monday, November 25, 2019 - link
Because Inteltech takes Intels $$$ and its one of the few areas where Intel doesn't get smashed.I agree with you, the main uses for these these kinds of CPU's are proper work not gaming. And definitely not gaming at 1080p. Its a joke.