A series of performance benchmarks for MD Apps, including GROMACS

[pszilard]

@Entropy_YU

By “theoretical capabilities” I mean the ranking of different GPUs in tests of other MD applications that directly utilizing the HIP codes (e.g., Amber and OpenMM in my tests). AMD GPUs rank significantly lower in GROMACS than in other applications. If we look at some 3D graphics benchmarks (e.g., 3DMark), we’ll also see that AMD and intel GPUs don’t rank like GROMACS. In my expectation, the 6900XT should have better performance than the 3080Ti, and the Intel ARC A770 should have a performance closer to the 4060Ti. While of course, this is all my personal subjective opinion.

A few points related to what you seem to be assuming:

• (synthetic) graphics benchmarks do not necessarily reflect MD performance;
• MD codes use different algorithms and implementations and are likely tuned to a different extent on different GPU architectures; an alternative explanation could be that some applications are less efficient on NVIDIA than on AMD; e.g. GROMACS is well tuned on NVIDIA and has been optimized on CDNA but not much effort has gone into RDNA optimizations;
• native HIP use equals better performance is not a given, algorithms, optimizations, “luck” with compilers often matter more; at least for our algorithms and implementation currently there is at most 10-15% performance left on the table vs plain HIP (on single GPU for large system sized) and that is on CDNA2 GPUs, I’m not sure about RDNA but I guess that number is smaller.

@Entropy_YU

Yes, I have noticed that some tests on the Internet (especially the NVIDIA’s tests) are not comparable between different applications because many of the parameters vary too much. So I used the same parameters to compare the “efficiency” of different applications in the 4090 test last year, which is also mentioned many times in today’s blog. You could look at the last picture in that post, where the 4 models (STMV-NPT, B, A-2 and A) have exactly the same parameters between applications.

Thanks for the pointer. The pictures don’t seem to load for me (or are very slow to load), will try again later. Do you have a PDF version of your post – that may be easier to load?

I would recommend using “application performance” or just “performance” instead of “efficiency” since the latter implies that you are using a relative metric (e.g. performance relative to the best achievable would be an “efficiency”).

Cheers,
Szilárd