Not just a visual eye-catcher – Corsair Vengeance RGB RT 2x 16 GB 3600 CL18 Kit Review with Overclocking and Teardown

1 - Introduction and specifications 2 - Unboxing and first impression 3 - Dimenstions and lighting 4 - SPD and heatsink performance 5 - Teardown and PCB analysis 6 - Test systems and methodology 7 - XMP behavior and overclocking 8 - Synthetic benchmarks – AIDA64 and Geekbench 3 9 - Gaming – Cyberpunk 2077 in UHD, QHD, FHD 10 - Final thoughts and conclusion

The behavior when loading the XMP profile is flawless and without any further intervention in voltages or timings, the kit runs at 1800 MHz in 1:1 mode with our 5950X. While the secondary timings are only partially adopted, which should provide a noticeable performance disadvantage especially with tRC, this is not the fault of the RAM kit, but of the board or BIOS that makes these decisions.

But even so, you’re already getting a lot of the performance out of your AMD Ryzen platform with just a single BIOS “click”, not least because of the dual-rank topology of the modules, where the ranks can effectively take turns doing work, improving efficiency.

If you now only increase the clock rate from DDR4-3600 to DDR4-3800, you already have to pay attention to the Infinity Fabric clock rate on many motherboards and CPUs if you don’t want to slip into the suboptimal 1:2 mode. Although AMD initially advertised the Ryzen 5000 generation with up to 2000 MHz FCLK, this only really materialized in extremely rare cases – keyword WHEA error – and accordingly many motherboard manufacturers are conservative with “auto” settings for the FCLK clock.

So long story short, the second config manually set FCLK to 1900 MHz, UCLK:MCLK mode to 1:1, VSOC to 1.15V, VDDG CCD to 1.0 V, and VDDG IOD to 1.1 V. Strictly speaking, the RAM still runs with “auto” timings, but the CPU already has to be set up a bit more precisely.

In the third config (*) I manually optimized all timings as seen above to squeeze the real performance maximum out of the Vengeance RGB RT modules. The primary timings are still relatively loose at 18-22-21-40, but the secondary and especially the tertiary timings could be tightened up a good bit, which should provide another noticeable performance boost. This requires some effort, but it’s not witchcraft and ultimately just trial-and-error with a RAM stress test like TestMem5.

DDR4-4000 could still be set, but really only the fewest CPUs can keep up with their FCLK clock here and in 1:2 mode you simply lose performance. Also higher clock rates are not possible with this kit, even with higher voltages or looser timings. On the contrary, above 1.4V the Corsair kit actually becomes more unstable, resulting in bluescreens, which is another indication that the ICs are Samsung 8 Gbit C-Die. On the positive side, however, the 1.35 V from the XMP profile was still completely sufficient for stable operation for the most performant overclocking.