The Big Gun: AMD Ryzen Threadripper 3990X Workstation Performance Review

While this article has no lack of synthetic benchmarks, SiSoftware’s Sandra makes it very easy to get reliable performance information on key metrics, such as arithmetic, multimedia, cryptography, and memory. Sandra is designed in such a way that it takes the best advantage of any architecture it’s given, so each CPU always has its best chance to shine.

That means a couple of things. This is definitely the “best” possible performance outlook for any chip, and doesn’t necessary correlate with real-world performance in other tests. It’s best used as a gauge of what’s possible, and to see where one architecture obviously differs from another.

Multimedia

Many tests in Sandra are designed to eke as much juice out of a processor as possible, and when more cores are involved, it’s almost guaranteed that tests like multimedia and arithmetic will see obvious benefits. That’s the case here with the multimedia test, with the 64-core monster flying 60% ahead of the 32-core. This is one test where AVX-512 can come into play, which helps Intel not fall further behind than it is. Now imagine if team blue had a 64 core!

Arithmetic

We’ve gone from a 60% improvement in the multimedia test to a 75% one in this arithmetic test. AVX-512 is not involved here, so the 3990X has less of a fight to battle. Clearly, if you need quality CPU compute performance, you’re not going to get it from a smaller chip.

Cryptography

Synthetic tests tend to push hardware really well, but this cryptography test is an exception where the 3990X is concerned. AMD’s top chip performs a bit better with the higher security test, but with Intel’s AVX-512 being utilized across the board, it doesn’t look as impressive as it could be. That bodes well for Intel’s 18-core competitor.

Memory Bandwidth

Memory bandwidth across AMD’s and Intel’s quad-channel platforms is pretty similar across-the-board, although we monitored less bandwidth overall with the 3990X than the other Threadrippers. It still comes far ahead of the dual-channel platforms. 40GB/s+ is really great to see there, thanks to the fast memory we’re able to reliably use nowadays.