How this calculator works
Double data rate memory sends two transfers per memory clock cycle, so a kit sold as DDR4-3200 actually runs its memory clock at 1600 MHz. CAS latency counts how many of those clock cycles pass between the read command and the first data. Multiplying the cycle count by the clock period gives the delay in nanoseconds, which is the only figure that compares fairly across speeds. A DDR5-6000 CL30 kit and a DDR4-3200 CL16 kit both land at 10 ns, even though one timing number is nearly twice the other. Peak bandwidth is separate: it depends only on the data rate and the channel width.
Formula
true latency (ns) = CAS latency (cycles) x 2000 / data rate (MT/s). CAS latency = true latency x data rate / 2000. data rate (MT/s) = CAS latency x 2000 / true latency. Memory clock period (ns) = 2000 / data rate. Peak bandwidth per 64-bit channel (GB/s) = data rate x 8 / 1000.
Worked example
A DDR4-3200 CL16 kit: 16 x 2000 / 3200 = 10 ns. The clock period is 2000 / 3200 = 0.625 ns, and the peak bandwidth of one 64-bit channel is 3200 x 8 / 1000 = 25.6 GB/s. A DDR5-6000 CL30 kit: 30 x 2000 / 6000 = 10 ns as well. If you want 8 ns from a 6000 MT/s kit, you need CAS latency 8 x 6000 / 2000 = 24 cycles.
Advertisement
Frequently asked questions
Why is DDR5 CAS latency so much higher than DDR4?
CAS latency is counted in clock cycles, and DDR5 kits run their memory clock much faster. A higher cycle count can still mean the same or a shorter delay. Compare true latency in nanoseconds, not the cycle figure, when you move between generations.
Does lower true latency always mean a faster computer?
No. Games and large data sets often care more about bandwidth and about how the memory controller schedules requests. Latency matters most for pointer chasing and small random reads. Capacity and channel count usually matter more than a one or two nanosecond difference.
Is the advertised data rate the same as the clock speed?
No. The advertised figure is the transfer rate in MT/s, which is twice the memory clock for DDR. Task Manager may show the clock in MHz, so a 3200 MT/s kit can appear as 1600 MHz. This page uses the transfer rate, matching the label on the kit.
Does the calculation include the rest of the timing set?
No. Only CAS latency is in this formula. The other numbers, such as tRCD and tRP, add their own delays but are not part of the first-word CAS latency. Use the full timing set when you compare kits at the same data rate.
What is first word latency?
It is another name for the same true latency: the time from a read command to the first word of data. Some tools call it absolute latency. It does not include the time to fill the rest of a burst.