Performance
Performance has been tested on the AMD platform, which contains a Ryzen 9 5900X processor, ASUS Strix B550-XE Gaming motherboard, ASUS Strix RTX3070 8GB graphics card, Silicon Power US70 1TB NVMe SSD, and Sharkoon 850W 80+ Gold PSU. Everything closed in the Sharkoon Elite Shark CA700 case.
For the first time, performance has been tested on Windows 11 Pro x64. Results were made before the L3 cache fix. However, all results were made using the ADATA Spectrix D50 ROG 16GB DDR4-3600 memory kit, so they are easy to compare. Used benchmarks were not much affected by the L3 cache issues.
All overclocking results were stable and passed multiple tests, as listed in the comparison.
As usual, we will start with the AIDA64 Memory and Cache benchmark, which is probably the best application to check memory bandwidth and latency.
AIDA64 bandwidth results at XMP settings are as high as expected. The overclocked results suggest how much more we can improve the performance if we have a proper motherboard.
The latency is also pretty good. We have seen better results, but it doesn’t mean that the tested ADATA memory kit is far behind. As we can see, results at a 1:1 memory controller to Infinity Fabric ratio, so up to DDR4-3800, offer us a much lower latency. As we can see on the above graph, the same results give us lower memory bandwidth. The final performance gain depends on the used application as some react better to latency and others to bandwidth.
In PCMark 10, the best results are at DDR4-3800 CL16. The DDR4-4600 CL18 result is also pretty good, but we can see that the DDR4-4400 is already losing a bit in most tests. In this comparison, the XMP setting seems like the easy to set and fast option.
3DMark series benchmarks are showing similar results in all settings. We would see a higher difference if we set RAM at a much higher frequency or higher frequency and significantly tighter timings. Because of the used platform, we can’t see significant differences in results even though the memory frequency difference is relatively high.
In VRMark, we can see up to 6FPS difference between the settings. The average FPS is not much different in the Cyan and Blue presets.
In Final Fantasy XV and Superposition benchmarks, we can again barely see any difference. Regardless of RAM settings, in these two tests, the results are similar.
The highest FPS difference is usually in popular games. However, in this case, it’s one more time almost the same. We can see 1-3FPS more or less, but it wouldn’t affect our gaming experience at all.
The quick conclusion from the performance tests is that the XMP profile is fast enough for everything and delivers high performance. Overclocking is not required and, at least on our test platform, is barely helping, so I suggest saving the time needed for tests and sticking with the XMP profile enjoying favorite games.
On the next page, I will share more info about overclocking for users who still wish to gain more than the XMP settings.
