This is a condensed adaptation of the original KeyboardTester.click buying guide: DDR5 CL30 vs CL36: Real Latency Math and When to Pay More.
Most DDR5 buying advice about CAS latency was written before the 2026 memory shortage. I re-checked what the CL30-vs-CL36 decision actually costs today, on live retail listings, and the old "$15–25 premium" rule of thumb is dead.
The whole formula, once
True latency in nanoseconds:
ns = 2000 x CL / MT/s
That is the entire secret. At DDR5-6000:
- CL30 → 10.0 ns
- CL36 → 12.0 ns
You are deciding whether 2 nanoseconds is worth the price gap. You can run the same math interactively (including reverse-solving the CL a faster kit needs to tie) in the free RAM Latency Calculator.
What the price gap looks like right now (Newegg, Aug 11, 2026)
| Kit (2x16GB DDR5-6000) | Price | True latency |
|---|---|---|
| G.Skill Flare X5 CL36 | $479.99 | 12.0 ns |
| G.Skill Flare X5 CL30 | $599.99 | 10.0 ns |
| Patriot Viper Elite 5 CL30 | $429.99 | 10.0 ns |
| Kingston FURY Beast CL30 | $649.99 | 10.0 ns |
Two things jump out:
- The same-brand CL30 premium is $120 — 25% of the kit price — for 2 ns.
- A cross-brand inversion: on the same day, a Patriot CL30 kit sold for less than the G.Skill CL36 kit. If you shop by nanoseconds across brands instead of walking one brand's ladder, the "premium" can be negative.
What 2 ns buys in games
There is no major independent lab head-to-head on exactly CL30 vs CL36 at 6000. The community consensus brackets it at roughly 1–4% in CPU-bound games, mostly showing up as slightly smoother 1% lows. A Tom's Hardware forum regular put it honestly: many times you cannot tell without a frame counter on screen.
Two anchors from actual test data:
- TechSpot measured a Ryzen 7700X gaining ~17% average FPS going from DDR5-5200 CL40 to 6000 CL30 — but that is speed and latency together, not the CL30/CL36 delta.
- On X3D CPUs the cache absorbs most of it: AMD's own numbers show about 1% average FPS (max 2%) across 30+ games from DDR5-6000 vs 4800 on the 9850X3D.
The free fix worth more than CL30
Every DDR5 kit boots at JEDEC defaults until you enable the EXPO/XMP profile. A "6000 CL30" kit running at its SPD default of 4800 CL40 is sitting at 16.7 ns — you paid for 10.0 and you are running 16.7. Toggling one BIOS setting is a bigger latency cut than any CL upgrade money can buy. (Check your kit's spec sheet: Kingston's FURY Beast 6000 CL30 lists exactly that 4800 40-39-39 default.)
The decision rule that survived the re-check
- Pay for CL30 when the day's real delta is small — roughly ≤10% of kit price.
- Shop across brands by nanoseconds, not by one brand's product ladder.
- Never replace a working CL36 kit chasing 2 ns.
- Enable EXPO/XMP before spending anything.
The full guide has the worked decision table, the shortage context (memory is now ~35% of a PC's build-materials cost, per HP's Q1 FY2026 earnings call), and all sources: DDR5 CL30 vs CL36: Real Latency Math and When to Pay More.
Top comments (0)