CAS Latency Explained: Why Lower CL Isn’t Always Better

CAS latency, written CL16 or CL30 on a spec sheet, is the number of clock cycles between the memory controller requesting a column of data and that data appearing on the bus. Because it is measured in cycles rather than nanoseconds, the same CL number means different real-world delays at different frequencies. Comparing CL figures across speeds without converting is the single most common mistake in memory shopping.

The conversion is simple. True latency in nanoseconds equals CL divided by half the rated transfer rate, multiplied by 2000. A DDR4-3200 CL16 kit works out to 10 ns. A DDR5-6000 CL30 kit also works out to 10 ns — identical first-word latency, with double the bandwidth behind it. A DDR5-6000 CL36 kit lands at 12 ns, which is why enthusiasts describe CL30 at 6000 as the current sweet spot rather than an exotic bin.

CL is also only the first of four primary timings. tRCD, the delay to activate a row, and tRP, the delay to close it, are usually equal to or larger than CL and are quoted second and third in the familiar 30-36-36-96 string. The fourth, tRAS, sets the minimum time a row stays open. A kit advertised as CL30 with loose tRCD and tRP can lose to a CL32 kit with tighter secondaries, which is why the full timing string matters more than the number on the box.

There is also a stability dimension. Kits binned for very low CL at high frequency need a strong memory controller in your CPU, and controller quality varies between individual chips of the same model. A CL30 DDR5-6000 kit is generally safe on current AMD and Intel platforms. A CL28 DDR5-7200 kit may need voltage tuning and may not post at rated speed on every board. If you do not intend to tune, buy the profile you know your platform runs.

The practical rule: convert to nanoseconds, then compare bandwidth. Buy the lowest true latency you can get at a frequency your platform officially validates, and ignore any comparison that puts a DDR4 CL number next to a DDR5 CL number as though they were the same unit.