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.
Results And Analysis
Retained hardware and software from our initial DDR4 round-up allows us to compare every DDR4 kit reviewed, without retesting. Almost as important is that I can show the link, rather than repost a bunch of redundant setup tables. We’ll jump straight to the point!
Intel’s higher memory multipliers are less stable than the more familiar validated settings, and the easiest way to get a more aggressive data rate with a lower memory multiplier is to use an elevated base clock. The CPU's 1.25x strap allows the parts that can’t be overclocked very far to run at 100MHz while the CPU core and memory controller operate at 125MHz. Memory multipliers that would normally correspond to DDR4-2400 and DDR4-2666 at 100MHz BCLK become DDR4-3000 and DDR4-3333 at 125MHz.
No memory we’ve reviewed thus far has been stable at a 3333 MT/s data rate, and our CPU becomes unstable at BCLK frequencies beyond 132MHz (corresponding to a 3168 MT/s data rate). Filling the hole between 3168 and 3333 requires underclocking the 125MHz BCLK, which would be silly if not for the fact that it exposes the true limits of each memory kit. In this case, Mushkin’s DDR4-2800 overclocks only one test setting lower than G.Skill’s highly-praised DDR4-3000.
Redline 994206F also overclocks one test setting higher than Corsair’s high-priced Dominator Platinum, though the price of those premium parts includes fancier heat spreaders and its Airflow Pro activity and temperature monitoring interface.
Manually configuring the Redline 994206F kit to its lowest stable primary timings, we see it mirrors Corsair’s Dominator Platinum, which has a similar DDR4-2800 rating. G.Skill's DDR4-3000 modules lead slightly.
Mushkin’s tiny lead over its DDR4-2800 rival continues through Sandra's Memory Bandwidth test, though only at default XMP frequency and timings.
The Mushkin difference is its default XMP latency. Corsair’s CAS 16-18-18-36 is designed for easy booting, while Mushkin’s 15-15-15-35 pushes performance hard right out of the gate. Since both kits support identical optimized values, I’d say that either Corsair was being cautious or Mushkin reckless. I’d love to hear your opinion in the response thread.
The Redline 99420F kit falls slightly behind under Grid 2’s most memory-sensitive settings, though I don’t know anyone who could see the actual difference between 231 and 234 FPS. Other benchmarks are mostly tied. I’d love to hear suggestions for another popular real-world application that I should add to the test suite.
A chart that shows actual performance for each memory kit, divided by the average performance of all four kits, divided by price, shows that Mushkin’s Redline 994206F achieves roughly the same value as G.Skill’s DDR4-3000. This happened only after G.Skill radically reduced the price of its premium kit. The current $5 delta raises the question of who I should recognize most prominently?
G.Skill’s recent price reduction sets up a conflict that I won’t be able to resolve to anyone’s satisfaction. Is it fair to award one product in the review of another? Would it be fair not to equally award products of equal standing? And how can I recommend two products to fill exactly one role in exactly one system? I’ll try to answer this, to no one’s satisfaction: We recommend that buyers choose between Mushkin 994206F and G.Skill 3000C15Q kits based on the current price at their favorite vendor, their favorite brand for support and their ability to configure custom settings.
