Why you can trust Tom's Hardware Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.
Test Results And Final Analysis
Mushkin’s DDR4-3200 CAS 20 could have a tough time competing against Teamgroup’s DDR4-3600 and G.Skill’s DDR4-3866 for bandwidth, but we’re really hoping it is also able to prove its worth in our price-to-performance analysis. On the plus side, it is also able to support lower-than-rated timings, though both competitors share that capability.
| Lowest Stable Timings at 1.35V (Max) on Asus Maximus IX Hero (BIOS 0801) | |||
|---|---|---|---|
| Row 0 - Cell 0 | DDR4-3200 | DDR4-3200 | DDR4-3200 |
| Mushkin RL Ridgeback 16GB MRB4U320LLLM8GX2 | 18-19-19-38 (1T) | 15-16-16-32 (1T) | 12-13-13-28 (1T) |
| T-FORCE XTREEM 16GB TXGD416G3600HC18ADC01 | 14-15-15-30 (1T) | 12-12-12-28 (1T) | 11-11-11-28 (1T) |
| G.Skill Trident Z 32GB F4-3866C18Q-32GTZ | 14-15-15-30 (1T) | 12-12-12-28 (1T) | 10-11-11-28 (1T) |
In the overclocking test, Mushkin’s DDR4-3200 failed the DDR4-3600 stability test after 15 minutes. Dropping the motherboard to a 44x CPU multiplier and 34.66x DRAM ratio, a 103 MHz BCLK brought the Ridgeback kit to a stable DDR4-3570. (For those interested, a 34.66x DRAM ratio uses a 13x memory multiplier with a 4/3 memory controller to CPU BCLK ratio. DDR doubles the data rate of the resulting clock rate.)
The above test also shows a maximum CAS Latency value of 21 cycles. While this did help the Mushkin Redline Ridgeback overclock a little higher, the Trident-Z modules worked better at CAS 19-21-21, hence the (Max) label.
Sandra Memory Bandwidth shows a preference for four-DIMM kits on Kaby Lake (as inherited from Skylake), and the difference has historically been around 10%. Even if we were to reduce the Trident-Z bandwidth numbers by 10%, we can plainly see how its better data rate and better timings would keep it in the lead. Completely removing that bias, the T-Force Xtreme kit has a smaller lead.
Two modules often beat four in Sandra Memory Latency, and we see the T-Force Extreme with the least latency at optimized settings. The Redline Ridgeback kit also has two modules, but its lengthier latency settings put it behind in this measurement.
Our overclocked Core i7-7700K and GTX 1080 provide enough performance that we can finally see the deficits of slower RAM in a racing game, without relying upon low graphics setting that a serious gamer wouldn’t use. The Redline Ridgeback falls behind the T-Force Extreme.
Metro Last Light Redux shows less bias for memory bandwidth or timings, and the T-Force Xtreme DDR4-3600 barely beats the Redline Ridgeback DDR4-3200.
Blender’s CPU render barely responds to improved memory performance, and both the Mushkin DDR4-3200 and Teamgroup DDR4-3600 memory complete the render operation at three minutes, one second, using XMP settings.
File compression applications often benefit from improved memory performance, and the scaling in 7-Zip is quite impressive . . . for everything but the Redline Ridgeback kit.
Redline Ridgeback DDR4-3200 CAS 20 is a little pricey at around $170, which is comparable to the DDR4-3200 CAS 16 of several competitors. Even at its most recently spotted $180, Teamgroup’s DDR4-3600 barely costs a little more.
The last chart, Performance Per Gigabyte, primarily shows that G.Skill’s 4x 8GB set has enough performance to overcome its $30 price premium compared to two of Mushkin’s 2x 8GB sets.
MORE: Best Memory
MORE: DDR DRAM FAQs And Troubleshooting Guide
MORE: All Memory Content
