Benchmark Results & Final Analysis
Since latency is a bad thing, we tried finding the lowest stable timing set for each module kit. Our motherboard couldn’t configure the Viper LED kit to workable DDR4-2666 or DDR4-2400 advanced timings, so we’ve excluded those from today’s benchmark charts. This has proven to be a firmware limitation, since it previously applied to some memory from another brand.
|Lowest Stable Timings at 1.35V (Max) on MSI Z370 Godlike Gaming (BIOS A.30)|
|G.Skill Sniper X 16GB F4-3600C19D-16GSXF||19-20-20-40 (2T)||16-17-17-34 (1T)||13-14-14-28 (1T)||12-13-13-28 (1T)|
|Patriot Viper LED 16GB PVLW416G360C6K||18-19-19-38 (2T)||15-16-16-32 (1T)||X||X|
|G.Skill Ripjaws V 16GB F4-3200C16D-16GVKB||X||15-16-16-32 (1T)||13-13-13-28 (1T)||11-11-11-28 (1T)|
|T-Force Dark ROG 16GB TDRRD416G3000HC16CDC||X||16-17-17-34 (1T)||13-14-14-28 (1T)||12-13-13-28 (1T)|
The Sniper X DDR4-3600 is the first RAM we’ve tested that overclocked higher using 21-21-21-42 primary timings, as opposed to 19-21-21-42. The difference was only a few MHz, as the same kit reached 19-20-20-40 timings at DDR4-3733.
Sandra Memory Bandwidth shows the mid-priced Sniper X trailing the higher-priced Viper LED at all settings, which makes sense since it requires slower primary timings. The barely-still-available dual-rank kit edges out both, when manually configured to optimal performance settings.
Lower is better when it comes to latency, and the Sniper X falls slightly behind the Viper LED there, as well. As an aside, the dual-rank kit started off with poor timings but tuned up nicely.
F1 2015 has unusually high memory bottlenecks compared to games that aren’t based on the EGO game engine, and it awards the Viper LED’s high data rate and good timings with a big lead over other single-rank modules. Even the dual-rank kit requires manual tuning to beat it. Could this test be the Sniper X’s Achilles’ heel?
As with most other games, Metro Last Light Redux has a fairly low memory performance requirement, after which other factors such as the graphics card are most important. The Sniper X loss is trivial.
Blender CPU Render is typical of most non-gaming apps, in that it too has a fairly low memory performance requirement, after which the CPU becomes the bottleneck. All compared kits satisfy its needs.
Remembering that less completion time means more performance, long bars on the Sniper X in 7-Zip mean that it loses noticeably. Even the Ripjaws V DDR4-3200 could be tweaked to produce lower completion time.
Price is where Sniper X takes its victory dance on Viper LED, its 11% lower cost offsetting its 6% lower performance to give it a margin in value. If you’re still looking at the dual-rank kit as a possible alternative, it might be gone.
And so the Sniper X takes the value edge over Viper LED, and the only things buyers must question is whether their specific apps are among the few that can benefit from the Viper LED’s performance lead, and whether or not they’re willing to also pay for the Viper LED’s lighting.
MORE: Best Memory
MORE: All Memory Content