Performance
Performance has been tested on two platforms. The first – marked in graphs in red, is AMD Ryzen 5 PRO 4650G. The second one – marked in graphs in blue, is AMD Ryzen 9 5900X. Both processors work on the ASUS Strix B550-I Gaming motherboard and ASUS TUF RTX3070 OC graphics card. All tests were performed on Windows 10 Pro x64 installed on Crucial P5 500GB SSD.
Our comparison includes overclocking results and settings at which the Ballistix MAX DDR4-4400 was stable. The maximum stable frequency for the Ryzen 5 4650G platform was 2650MHz or DDR4-5300. Still, because of higher than recommended voltages and the same results as at DDR4-5200, I’ve decided to use only DDR4-5200 CL18-24-22 1.55V setting in the comparison. Ryzen 9 5900X has a much weaker memory controller, and the maximum settings were DDR4-5000 CL18-22-22 1.55V.
As usual, we will start with AIDA64 Cache and Memory benchmark, which is probably the best software for synthetic memory speed tests.
As we can see, each processor has its own preferred settings, at which it works the best. However, in synthetic bandwidth tests, the higher frequency usually means much more than memory timings. Because of the CPU architecture, we can see a higher memory read on Ryzen 5900X processor, while the memory writes always looks better on Ryzen 4650G. The most important in daily work is the memory copy test, which is significantly higher on Ryzen 5900X.
Memory latency is important in overall performance. Here we can see that XMP settings on Ryzen 5900X are giving us around 67.7ns latency, which after additional adjustments can go below 60ns. This is a huge difference compared to Ryzen 3000 series processors, or even Ryzen 4000, where most results are closer to 80ns.
PCMark 10 is showing us high performance in mixed load tests. The best results are at lower frequencies, where timings are quite tight, and next at the highest frequencies, where is a quite good balance between memory latency and bandwidth.
In Cinebench series benchmarks, results are similar. These tests are not using much data, so results are based more on the CPU speed. The XMP looks well in these tests.
It’s time for some 3D benchmarks from UL(previously Futuremark).
3DMark benchmarks are showing us performance differences, mostly at lower graphics details. The Fire Strike test suggests that all results are similar on the Ryzen 4650G, while on the Ryzen 5900X, the best performing is DDR4-3733 CL14 and DDR4-5000 CL18. This is how much higher frequency we have to set to match the infinity fabric and memory controller ratio differences.
More demanding 3D tests at the display resolution up to 8K are not much different. We can see that mostly Final Fantasy XV benchmark reacts to memory settings, but both Superposition tests are almost the same in all cases.
Finally, something that likes fast memory. Modern games are often showing up to 10% performance gain because of memory settings. We can see closer to a 5% difference in our tests, but it’s still significant, especially that we are using the same memory kit for all settings.
I’m sure that everyone will be satisfied with the Ballistix MAX XMP profile. The memory runs fast at its declared settings what’s the most important. Additional overclocking is a bonus, which we are glad to see, especially that the memory is not cheap, so we look at something premium.
The overclocking range is vast for the Ballistix MAX memory, and to cover it, a separate article would be needed. However, I hope that some of our settings will be enough to convince you how great is this memory kit.
Let’s move to the next page, which will cover the overclocking topic.
