Performance
Performance has been tested on the Intel platform that contains the i7-14700K CPU, ASUS Z790 APEX motherboard, Colorful RTX4080 Advanced OC graphics card, and Deepcool PX1000P 80+ Platinum PSU.
All tests were performed on the Crucial Pro OC 32GB DDR5-6000 memory kit. The 32GB memory kit could overclock up to DDR5-7000 and CL38-45-45 1.40V, which is a pretty good result and, so far, our best with Micron IC. The DDR5-7000 CL38 setting was stable in our tests. You can expect it from the memory kit as long as your motherboard and CPU’s memory controller can handle it.
Let’s begin the tests.
AIDA64 memory benchmark scales well with the memory frequency. We can see that we gain a couple more GB/s with each step. The results are pretty good, and at least at 6000-6400MT/s, it won’t be significantly better on the best competitive memory kits. At 7000MT/s, we could expect a bit higher bandwidth and lower latency, but overclocking is never guaranteed and maybe one of our readers will have better luck on a different setup.
The latency is not spectacular, but it gives us a solid 20ns improvement over regular memory modules running at JEDEC settings. If we pushed tREFI then we would see a couple of ns better results, but it’s never guaranteed to run stable.
The difference in synthetic bandwidth and latency tests does not always show the whole story, so let us look at other tests.
PCMark 10 Applications benchmark shows us differences in popular Microsoft Office. The most significant performance gains can be seen in Excel, where overclocking settings are faster than the XMP. It’s also the most demanding test. In everything else, all settings show similar results.
3DMark tests show barely any difference at overclocked settings, but if we take a closer look at scores, then it’s clear that higher frequency memory is faster. I don’t think it matters out of competitive benchmarking.
Since the new version of Cinebench was released recently, we also switched to the new 2024 version.
The new Cinebench reacts slightly better to RAM performance in rendering. It’s not significant, but we can tell which setting is the fastest, especially though this benchmark is quite long.
The result at DDR5-6000 XMP is already pretty good. The lower XMP profile is significantly slower.
Final Fantasy XV and Superposition results at high display resolutions are barely different. The difference between the slowest and fastest settings is not much higher than the error margin.
Modern 3D games react much better to RAM performance at lower display resolutions like 1080p. In Cyberpunk 2077, with disabled Ray Tracing and DLSS, we can see up to 6FPS between various memory settings. In some other games, like the already old Shadow of the Tomb Raider, it can be even as high as 30FPS. On the other hand, higher display resolutions at higher details rely more on graphics card performance, and then we can see significantly lower gains – about 2FPS in titles like Assassin’s Creed Valhalla. Surprisingly, FarCry 6 shows a 6 FPS gain going from the 5600MT/s XMP to the 7000MT/s OC profile.
The XMP 6000 profile is reasonably balanced between standard and highly overclocked settings. The performance is pretty good, and even though it could be better, then it wouldn’t change the user’s experience.
On the next page, I will tell you some more about the overclocking of the new Crucial memory.
