Review – 32GB (2x16GB) Corsair Vengeance RGB DDR5 6400CL36 – Is a Micron Chip Kit a Good Choice?

In this review, we will analyze a DDR5 memory kit from Corsair’s Vengeance RGB series, which offers models ranging from 5200 to 8200 MT/s in modules of 8GB, 16GB, 24GB, 32GB, or 48GB. The kits are available for purchase individually or as part of larger bundles such as 16GB, 32GB, 48GB, 64GB, or 96GB. For this analysis, we tested a kit featuring two 16GB modules totaling 32GB, operating at 6400 MT/s with timings of 36-48-48-104 and a voltage requirement of 1.35V.

The memory modules come in a rectangular box featuring an illustration of the product, the manufacturer’s logo, and model number. On the back side, there are descriptions in multiple languages and windows where you can view the modules. The code for this memory is CMH32GX5M2B6400C36.

Regarding the heatsink, Corsair opted for a simple aluminum component that makes contact with the chips and PMIC. It does not feature fins, so it functions more as a “buffer” than a traditional heatsink, though at least the plate covers the entire area of the chips and PMIC.

The tested sample features RGB lighting and is compatible with Corsair’s iCue software, which allows synchronization and control with other RGB devices from the manufacturer.

Upon removing the heatsink, we were surprised to find chips provided by Spectek, a division of Micron that supplies memory chips and NANDs at a more affordable cost. Unfortunately, it was not possible to precisely identify which chip this is; however, the strongest suspicion is that they are the Micron D-Die. It’s worth noting that these are 16 Gbit dies and undoubtedly newer than the Micron A-Die models from DDR5 launch, which cannot operate at the standard specifications of these modules.

Regarding the PMIC, the integrated circuit responsible for powering the memory module, a Richtek unit was adopted.

Here you can view the information stored in the SPD of these memories, including the XMP profiles and other specifications.

Here is the link to the product page on the manufacturer’s website: https://www.corsair.com/br/pt/p/memory/cmh32gx5m2b6400c36/vengeance-rgb-32gb-2x16gb-ddr5-dram-6400mt-s-c36-memory-kit-black-cmh32gx5m2b6400c36#tab-downloads

Hardware used :

CPU: AMD Ryzen 7 8700G e Ryzen 7 9700X (Thanks AMD!)

MOBO: ROG Crosshair X670E Gene (BIOS: 2505)

RAM: 2×16 GB Corsair Vengeance RGB DDR5-6400CL36 1.35V – CMH32GX5M2B6400C36 – (Thanks Alexandre Pimenta!)

GPU: Powercolor RX 6700 XT Fighter 12 GB

PSU: Coolermaster MWE 1250 Gold V2 (Thanks Cooler Master)

COOLER: Watercooler Custom

SSD: Crucial BX300 120 GB + Teamgroup T-Force Vulcan Z 1TB

Software: Windows 10 22H2 x64, TM5 0.13.1 1usmus, Geekbench 3.4.4, y-cruncher 0.85, Counter Strike 2, Shadow of the Tomb Raider and 3dmark11.

Testing methodology

Discover the daily usage limits of Corsair memory modules using the Ryzen 7 8700G and 9700X processors, with the goal of exploring optimal settings for each CPU:

1. XMP/EXPO : The objective here is to test if it’s possible to operate with stability using the EXPO profile, if available, or the factory XMP setting, eliminating the traditional overclocking tests we used to perform under these conditions.

This change was necessary because “easy overclocking” no longer makes sense for DDR5 memory. The gains from doing so are practically non-existent since motherboards tend to relax subtimings significantly, negating any benefits from overclocking.

2. 24/7 with fine adjustments : Manual adjustments were made to all possible parameters to achieve the best results feasible for daily use, pushing the limits of 1:1 and 1:2 modes with the 9700X.

For cases 1 and 2, both CPUs were locked at 5.0GHz, with FCLK set at 2500MHz for the 8700G and 2200MHz for the 9700X. It’s important to note that for Ryzen AM5 processors, there’s no longer the need to maintain the FCLK running in proportion to memory clock. In other words, before adjusting these settings, it’s advisable to test the FCLK limit, which for Ryzen 7000 processors falls between 2000 and 2200 MHz, and for Ryzen 8000 processors, between 2200 and 2700 MHz.

It is worth highlighting that the adjustments obtained with the Ryzen 7 9700X can also be applied to Ryzen 7000 processors since both series share the same IO Die and therefore, the same memory controller.

Timings – AMD

XMP / EXPO:

For this model, Corsair included a DDR5-6400 XMP profile, which worked well on the test hardware, with the motherboard automatically applying UCLK in 1:2 mode at 6400 MT/s on the 8700G.

Again, it’s important to note that this is a limitation of this CPU sample, which may also apply to Ryzen 7000 and 9000 processors. In some cases, certain examples aren’t stable with UCLK running at 1:1 and memory operating at 6400 MT/s, making it more desirable to seek the maximum stability while maintaining the 1:1 ratio.

24/7 with fine-tuning:

When performing fine-tuning adjustments, it was possible to improve several timings and reach 7600 MT/s with both the R7 8700G and R7 9700X, representing a significant leap compared to the launch Micron A-Die modules, which, with a lot of luck, could only reach 5600 MT/s. It’s interesting to note that achieving this required reducing VDD and VDDQ from the standard 1.35V to 1.29V because higher voltages caused errors in TM5 testing. Despite this behavior seeming unusual, it’s not the first time we’ve seen memory performing better with lower voltage, as Samsung C-Die DDR4 modules can also exhibit this characteristic.

Regarding timings, unlike what typically happens with the well-established Hynix chips—where several parameters scale with additional voltage—it wasn’t the case with these Micron/Spectek modules. This is likely because these memories performed better with undervolting, despite that, the primary timings ended up more relaxed than what we usually see in Hynix A-Die modules. Additionally, tRFC exceeded 250ns even with the 8700G.

The key timings to relax for achieving higher frequencies were primarily the primary timings, especially tCL and tRCD, tRC, maintaining the SCL proportion of tWRWRSCL = 2*tRDRDSCL, tRFC, and also termination impedances.

Benchmarks

Moving forward, we have the performance numbers from the benchmarks, and it’s important to note that these results can vary across different processors and platforms. For example, higher memory frequency benefits are more noticeable on Intel 12th/13th/14th Gen CPUs, while on Ryzen 7000 and 9000 processors, the gaming performance gain tends to be negligible in 1:2 mode, even at 8000 MT/s.

Remember that all these results passed the TM5 0.13.1 stability test and, specifically for these samples, represent something that can be used daily without issues.

2D Benchmarks

In synthetic memory benchmarks, that’s where the difference from overclocking usually becomes more evident, with both Geekbench and y-cruncher 0.85 1b showing significant gains when overclocking is applied. However, these improvements don’t always translate to other 2D applications, as we saw in our optimization guide for the Ryzen 7 8700G.

The gains were more pronounced on the 8700G because it benefits from a higher FCLK, which ensures greater effective memory bandwidth in these scenarios. In contrast, the 9700X didn’t scale much from 6400 to 7600 MT/s.

3D Benchmarks

In addition to 2D tests, we also ran gaming benchmarks to assess performance gains from timing and frequency optimizations. The tests were conducted with the following configurations, aiming to simulate a scenario where the CPU is the limiting factor. This setup represents the best-case gain scenario for the tested CPUs, which tends to decrease with higher resolutions.

We also included 3DMark11, even though it’s an older tool. Its Physics and Combined Tests are highly dependent on the memory subsystem being well-tuned.

Results:

The memory overclock gains were more significant on the 8700G compared to the 9700X because the 8700G has half the L3 cache of the 9700X, leading it to perform more memory accesses. In this scenario, a faster memory subsystem and interface help compensate for this limitation.

With the 9700X, the bigger gains came from optimizing timings even at the default EXPO settings in 6400 MT/s, with little to no benefit from pushing up to 7600 MT/s. However, this allows for a lower VDDSOC voltage, freeing up some TDP headroom for the CPU.

Conclusion

The Corsair Vengeance RGB kit of 32 GB (2×16 GB) DDR5-6400CL36 showed good compatibility with the AMD platform, running smoothly with the XMP profile at 6400 MT/s without user intervention. However, this resulted in a UCLK at 1:2 mode with the 8700G, which can vary depending on the CPU and motherboard used and has little to do with the memory itself.

With manual adjustments, it was possible to achieve stable DDR5-7600 CL40 performance, which is below what kits using established Sk Hynix chips are capable of but represents a significant improvement over first-generation Micron chips, which barely reached 5600 MT/s. With the Ryzen 7 9700X, it was possible to reach DDR5-6400CL36 with optimized timings, which, while more relaxed compared to Hynix, can still be considered acceptable for daily use.

Regarding availability and price, the 32 GB Corsair Vengeance RGB 6400CL36 kit can be found around 99 USD , which is not a bad price for a 32GB kit. However, it’s possible to find models equipped with Hynix chips at almost the same price, which would be a better option in terms of performance. Therefore, for this Corsair kit to be a good option, it needs to be priced competitively.