Performance
Performance has been tested on the AMD Ryzen platform which includes the Ryzen 5 3600 processor and ASRock X570 Extreme4 motherboard. In 3D tests has been used MSI GTX1660Ti graphics card which isn’t the fastest so stresses other components some more.
Everything is readily available in stores so our readers can compare results on their computers.
All results were performed on the V-Color Prism II 16GB DDR4-3600 CL18 memory kit, so this is what you can expect but of course, overclocking is never guaranteed as it depends on many factors.
Let’s begin the tests.
Memory bandwidth in AIDA64 is going up to DDR4-3800 when it works at 1:1 divider with the infinity fabric. Above this frequency, results are going down and again up once the memory clock is higher. Memory Copy results are scaling with higher memory frequency. As we will find out later, results in AIDA64 are not showing real performance in popular applications and games but are still good for synthetic memory bandwidth comparison.
For those users who are interested in memory latency, I will add that at DDR4-3600 CL16 we can see about 68ns, at DDR4-3800 CL16 latency is about 66ns while at DDR4-4800 it’s 72ns. Because of the infinity fabric dividers, we can compare performance at DDR4-3800 to that of DDR4-4800. In most cases, users will be able to stabilize memory at DDR4-3600 so depends on the CPU and used motherboard, higher memory frequency can be pointless.
Rendering benchmarks like Cinebench R15 are showing slightly higher results at higher memory frequency but nothing really significant. I can say that the XMP setting is easy and safe option if we wish to keep high performance and won’t waste time on additional stability tests.
The PCMark 10 results are already starting to show some differences between memory settings. On the other hand, results are still so close that if we translated our scores to daily work then no one would feel the difference.
3DMark results are scaling pretty much the same as PCMark so higher memory clock gives slightly higher scores.
Results in VRMark are showing what I already said during AIDA64 tests. Memory is scaling rather with the infinity fabric clock than the memory frequency itself. Performance is going slightly up until we switch to 1:2 divider when it drops by 1FPS and next is going up. As we can see, the difference in all memory settings is +/- 1FPS. Nothing that will really affect the gameplay.
Results in Final Fantasy XV 2K and Superposition 8K benchmarks are showing a similar scenario. Even though faster memory settings are giving us some more points in these benchmarks, then if we translate it into FPS then we will see barely any improvement. Depends on the run, results are about 1-2FPS better at higher memory settings than the XMP.
Games like FarCry 5 or Shadow of the Tomb Raider are showing again the same improvements so about 1-2FPS more at higher memory settings. In this case, it’s no point to pay a premium for a faster memory kit. However, we can overclock our memory kit but for some users, it will be wasted time. I’m sure that players rather spend time in their favorite game than try to stabilize overclocked settings. The XMP profile is just perfect, even though at first sight, relaxed timings suggest lower performance.
On the next page, I will tell you some more about the overclocking of the V-Color Prism II 16GB DDR4-3600 memory kit.
