Any 3dmark experts?

[wyssn11]

Any 3dmark experts can help explain why 7975WX scores lower than 3975WX ?
2 product generation difference plus DDr5 vs DDr4



VS

 

[soreloser]

Because steel nomad is a GPU benchmark. Both of the CPUs will not matter. Go and run a test on cinebench R23 and you will see the big difference between the 2 threadrippers. Either that or run 3DMark timespy, i know theres a cpu benchmark in there but i'm not sure how well it will do for gaming scenarios.

 

[wyssn11]

Thank you for the explanation and recommendations.
Perhaps this particular benchmark is not able to reflect the full capabilities of the 7975WX.
Based on the 3dmark timespy results, it seems the gains are marginal or none at all.

3dmark timespy results:



VS

 

[soreloser]

You should use the average as the baseline in 3DMark, not the top score. This is because the top scorers are usually people who overclock etc and tune to the max to get their results, so its not very realistic. Cinebench R23 is a benchmark that focuses on the CPU only, so the results there should be accurate compared to 3dmark tests that do GPU and CPU, combining the results into overall score.

 

[matique]

Thank you for the explanation and recommendations.
Perhaps this particular benchmark is not able to reflect the full capabilities of the 7975WX.
Based on the 3dmark timespy results, it seems the gains are marginal or none at all.

3dmark timespy results:

Normal timespy does not utilise multicore as much as timespy extreme. Timespy is also heavily dependent on RAM OC. On timespy extreme, your chip should be better utilised and would take Singapore #1 spot from me Here's how the top few 7975wx fares in TSE, versus just the 15K CPU score of my 13900K. 7975WX is much faster in this bench.

On normal timespy, it would be difficult to beat the 13900K/14900K as it does not utilise much cores and is heavily dependent on RAM OC. My 13900K has less cores, but they're much faster cores paired with fast 8000c32 DDR5 RAM. Here's my normal timespy run:

I think your chip would also do very well in the CPU profile test. Take advantage of all your cores and you will probably get the highest score in all threads. Don't go for the heavy GPU based benchmarks.

Don't stress too much about benchmarks, they're not a very accurate way of judging performance. Like others mention, top scores are configured in such a way that they get the best scores, and may not be reflective of real world use. Nevertheless, have fun benching!

 

[SkyShroud]

I prefer to use geekbench for CPU benchmarking.