Performance Tests
Performance test at declared DDR4-3333 settings were performed on Intel Skylake platform which contains i7 6700K processor and MSI Z170I Gaming Pro AC motherboard with the latest BIOS/UEFI 1.91 which was in the beta version during our tests while now should be available its official version.
Additionally for 3D tests has been used ASUS GTX980 Strix graphics card and all was working in Win 10 Pro x64 installed on Crucial MX300 SSD.
Nothing unusual and I assume that many our readers have similar setup at home. The only thing which is worth to add is that processor’s cores and cache have been set to 4.2GHz. It’s not hard to do on any unlocked Skylake processor while it’s better showing any differences in memory performance between memory kits.
First as usual is AIDA64 Memory and Cache benchmark which is the best this type application to measure memory and cache bandwidth.
HyperX Predator performs great in this test what was expected as Intel Skylake simply loves high memory frequency. Results above 45GB/s are more than any gamer needs but of course more is better!
HyperPi 32M is showing pure calculations. This benchmark is scalling great with fast memory. As we see, not always higher frequency is better as Crucial memory at DDR4-2400 is showing better performance than some other, higher frequency kits. However HyperX Predator DDR4-3333 is still faster.
Rendering benchmarks are showing lower differences between memory kits as performance of any DDR4 memory is already high. We can still see that with higher frequency performance is better. I guess that in this case it counts only for all who are rendering for long hours so mainly professional graphics.
3DMark is also not showing big differences between memory kits as memory performance affects mainly physics test in this benchmark. It’s one more benchmark where we see that regardless of speed, performance is good.
About the same as in #DMark we can see in PCMark 8. Higher memory frequency is not helping much. All memory kits in our comparison perform good while HyperX Predator achives the best results.
If you still wish higher performance then there is always some overclocking headroom. Let’s take a look how HyperX Predator DDR4-3333 memory is overclocking.
Overclocking
Overclocking is never guaranteed so the presented results may vary from results on other memory kits. I am not recommending overclocking if you do not know what are you doing. High voltages may damage hardware and it will not be covered by warranty.
As a test platform for overclocking has been used MSI Z170I Gaming Pro AC motherboard and i5 6600K processor.
On this platform we were able to overclock XyperX Predator DDR4-3333 to DDR4-3466 at the same XMP settings so 16-18-18 main timings and memory voltage of 1.35V. Regardless of voltage or timings we couldn’t set any higher frequency which could be at the same time stable … or at least pass HyperPi 32M test.
Maximum at 1.45V was DDR4-3600 which was also working the best at CL16-18-18. At these settings memory could pass some tests but wasn’t fully stable. Below screenshot is showing expected memory bandwidth in case if you were lucky with your memory kit.
Overclocking is never guaranteed so any single MHz above what manufacturer is guaranteing is good. We were counting on some more but DDR4-3466 is still good result. Even without overclocking performance is great so there is no need to set it at higher frequency at all cost.
2 comments
This is making me want to take out my EVGA memory based on the same 8Gbit Hynix ICs and do some testing, as they are quite neglected.
I don’t know what about your kit but my EVGA kit which was based on older Hynix batch couldn’t pass ~DDR4-3200 stable. Actually this HyperX Predator DDR4-3333 is the only Hynix kit which I was testing that could reach DDR4-3600 without issues. Most others couldn’t run at more than 3200/3333.