Memory kits are sold with two headline numbers: the speed, such as DDR5-6000, and the CAS latency, such as CL30. Faster speeds look better, but the CAS latency is counted in clock cycles, not nanoseconds, so a higher CL on a faster kit does not necessarily mean slower memory. To compare kits fairly you need the true latency in nanoseconds.
The “speed” is really a data rate in megatransfers per second (MT/s). DDR memory transfers data twice per clock cycle, so DDR4-3600 runs its clock at 1,800 MHz, and one clock cycle lasts 1 ÷ 1,800 MHz = 0.556 ns. Multiply by the CAS latency to get the delay between the memory controller asking for a column and the first data arriving: 16 × 0.556 = 8.89 ns. That is why the formula is CL × 2000 ÷ MT/s.
Over the generations, data rates have climbed while true latency has stayed around 9–15 ns: DDR5-6000 CL30 has 10 ns latency but far more bandwidth than DDR4-3600 CL16. For gaming and latency-sensitive work, a lower ns figure helps; for bandwidth-hungry tasks, the higher data rate usually wins. The calculator gives both and compares two kits side by side.
How to use the RAM Latency Calculator
- Data rate (e.g. DDR4-3600 → 3600)
- CAS latency (CL)
- Compare with a second kit?
- Kit B data rate
- Kit B CAS latency
Results update as you type, every field has its own unit menu, and everything is calculated in your browser.
How it works
Enter the data rate (or the real clock in MHz) and the CAS latency, and optionally a second kit to compare. The calculator divides 2000 by the data rate to get the cycle time in ns and multiplies it by CL.
Worked example
DDR4-3600 CL16: the clock runs at 1,800 MHz, a cycle lasts 2000 ÷ 3600 = 0.556 ns, and the latency is 16 × 0.556 = 8.89 ns. DDR5-6000 CL30 gives 30 × 2000 ÷ 6000 = 10.0 ns, 1.11 ns more, but its bandwidth of 48 GB/s per channel is two-thirds higher than DDR4-3600’s 28.8 GB/s. Among DDR5-4800 kits, CL40 gives 16.67 ns and CL38 gives 15.83 ns.
Typical kits
| Kit | Latency | Bandwidth per channel |
|---|---|---|
| DDR3-1600 CL9 | 11.25 ns | 12.8 GB/s |
| DDR4-3200 CL16 | 10.00 ns | 25.6 GB/s |
| DDR4-3600 CL16 | 8.89 ns | 28.8 GB/s |
| DDR5-4800 CL40 | 16.67 ns | 38.4 GB/s |
| DDR5-6000 CL30 | 10.00 ns | 48.0 GB/s |
Inputs explained
| Field | What to enter |
|---|---|
| Data rate (e.g. DDR4-3600 → 3600) | Units: MT/s, MHz real clock (DDR) |
| CAS latency (CL) | A number |
| Compare with a second kit? | Choose one: No, Yes |
| Kit B data rate | Units: MT/s, MHz real clock (DDR) |
| Kit B CAS latency | A number |
What each result means
- True (first-word) latency = 2000×[CAS latency (CL)]÷[Data rate (e.g. DDR4-3600 → 3600)]
- Real clock frequency = [Data rate (e.g. DDR4-3600 → 3600)]÷2
- Clock cycle time = 2000÷[Data rate (e.g. DDR4-3600 → 3600)]
- Peak bandwidth per 64-bit channel = [Data rate (e.g. DDR4-3600 → 3600)]×8÷1000
- Kit B latency = 2000×[Kit B CAS latency]÷[Kit B data rate]
- Kit B bandwidth per channel = [Kit B data rate]×8÷1000
- Latency difference (B − A) = [Kit B latency] – [True (first-word) latency]
Formula
Latency (ns) = CL × 2000 ÷ data rate (MT/s) · real clock (MHz) = data rate ÷ 2 · cycle time = 2000 ÷ data rate · bandwidth per 64-bit channel = data rate × 8 bytes
Worked example
Inputs — Data rate (e.g. DDR4-3600 → 3600): 3,600 MT/s; CAS latency (CL): 16; Compare with a second kit?: Yes; Kit B data rate: 6,000 MT/s; Kit B CAS latency: 30.
Results — True (first-word) latency = 8.889 ns; Real clock frequency = 1,800 MHz; Clock cycle time = 0.5556 ns; Peak bandwidth per 64-bit channel = 28.8 GB/s; Kit B latency = 10 ns; Kit B bandwidth per channel = 48 GB/s; Latency difference (B − A) = 1.111 ns; Lower latency = Kit A.
Key terms
- CAS latency (CL)
- Clock cycles between a column read command and the first data.
- MT/s
- Megatransfers per second – the effective data rate of DDR memory.
- Real clock
- The memory clock frequency, half the DDR data rate.
- First-word latency
- CAS latency expressed in nanoseconds.
Frequently asked questions
How do I calculate RAM latency in ns?
Multiply CAS latency by 2000 and divide by the data rate: DDR4-3200 CL16 = 16 × 2000 ÷ 3200 = 10 ns.
Is DDR5 slower than DDR4?
Not really – DDR5-6000 CL30 has 10 ns latency, similar to DDR4-3200 CL16, with much more bandwidth.
Is lower CAS latency always better?
Only at the same data rate. Compare true latency in ns across different speeds.
Why is DDR4-3600 called 1800 MHz?
DDR transfers data twice per clock, so 3,600 MT/s needs a 1,800 MHz clock.
What is a good RAM latency for gaming?
Around 10 ns or less is considered fast; 8–9 ns kits are premium.
Does CAS latency include all memory delay?
No – it is the column access time; total latency also includes tRCD, tRP and the memory controller, typically 60–90 ns overall.