Saved

Press the star next to a tool's name to save it.

All my tools

RAM Latency Calculator

Use this RAM latency calculator to convert CAS latency to nanoseconds from the memory speed in MT/s, add tRCD and tRP, or compare two DDR memory kits.

Restored your last values ·

MT/s
cycles
cycles
cycles
MT/s
cycles

Enter the speed of your memory in MT/s, as in the product name, such as 3200 for DDR4-3200, and its CAS latency. You get the latency in nanoseconds, which lets you compare modules of different generations. Add tRCD and tRP to see the other access times.

Everything is worked out in your browser. Nothing you type is sent anywhere.

–

–

Enter the memory speed and the CAS latency, then press Calculate.

See full result

A RAM latency calculator converts the CAS latency of memory from clock cycles into real time, in nanoseconds. A kit labelled DDR4-3200 CL16 and a kit labelled DDR5-6000 CL30 look very different, but the calculator shows that both take 10 nanoseconds to start delivering data. Enter the speed and the CAS latency to get the true latency, or enter two kits to compare them. It also gives the clock rate and the other access times if you add tRCD and tRP. It runs in your browser.

How to use the RAM latency calculator

  1. Enter the memory speed in MT/s, such as 3200.
  2. Enter the CAS latency, the number after CL.
  3. Add tRCD and tRP if you want the other access times, or choose compare to test two kits.
  4. Press Calculate.

What CAS latency is

The Wikipedia article on CAS latency defines it as the delay in clock cycles between the READ command and the moment that the data is available.[1] It is counted in cycles of the memory clock, so the same number means a different time at a different speed. The page also warns that, for double data rate memory, you must use the clock rate, which is half of the transfer rate, when you work out the time.[1] DDR memory makes two transfers in each clock cycle.[1]

The formula

The pages do not give a formula for the time, so we derived it from those statements. A clock at half the transfer rate has a cycle that lasts 2,000 divided by the transfer rate in MT/s, in nanoseconds. The CAS latency in time is the number of cycles times that, so ns = CL × 2,000 ÷ MT/s. For DDR4-3200, the clock is 1,600 MHz, a cycle is 0.625 nanoseconds, and CL16 gives 16 × 0.625 = 10 nanoseconds. The calculator shows the clock, the cycle and the latency.

Examples that you can check

The CAS latency article has a table of the time for the first word of data. DDR3-1600 with CL9 takes 11.25 nanoseconds, with a cycle of 1.25 ns. DDR4-3200 with CL16 takes 10 nanoseconds. DDR5-6000 with CL30 takes 10 nanoseconds, DDR5-6400 with CL32 takes 10 nanoseconds, and DDR5-4800 with CL40 takes 16.67 nanoseconds.[1] The page that covers memory timings has the examples of DDR3-2000 with CL7, which is 7 nanoseconds, and DDR3-2666 with CL9, which is 6.75 nanoseconds.[2] Enter these into the calculator and it gives the same figures. The page has tags that say it needs citations, so treat the table as a guide.[1]

Why a higher CL can still be faster

The comparison mode shows why the number alone is misleading. DDR5-6400 CL32 has a higher CL than DDR4-3200 CL16, yet both take 10 nanoseconds, because the DDR5 clock is twice as fast. The transfer rate, and so the peak bandwidth, is double. If you compare DDR5-6400 CL32 with DDR5-4800 CL40, the first takes 10 nanoseconds and the second 16.67, so the first has a 40 percent lower latency and a third more speed. The calculator shows the percentage and the speed ratio.

Other timings

The article on memory timings explains that the timings are given in clock cycles, often written as four numbers such as 7-8-8-24, for CL, tRCD, tRP and tRAS. It says that the time to read the first bit from a bank with no row open is tRCD + CL, and that when the wrong row is open it is tRP + tRCD + CL.[2] The calculator turns these into time. For DDR4-3200 with CL16, tRCD 18 and tRP 18, the two cases are 21.25 and 32.5 nanoseconds. They show that a read that has to change rows takes about three times as long as the CAS latency alone.

Reading a memory label

A label like DDR4-3200 CL16-18-18-38 gives the transfer rate and then the timings, so the first number is the CAS latency and the next two are tRCD and tRP. Sometimes the label gives the clock instead, such as PC4-25600, which is the bandwidth in megabytes per second. For this calculator use the number after DDR4 or DDR5. If the box shows only the MHz clock, double it to get the MT/s.

Using the result when you buy

When you choose memory, compare the nanoseconds of the CAS latency and also the speed. For most users, a kit with a latency near the best of its generation and a high speed is a good choice. The very top kits cost much more for small gains, and a program that is limited by the processor, not the memory, will not benefit. If you overclock, compare the latency before and after, because raising the speed can force a looser timing that cancels the gain.

Common mistakes

  • Using MHz as MT/s. The clock is half of the transfer rate.
  • Comparing CL across generations. Compare the nanoseconds.
  • Ignoring bandwidth. Higher speed also raises the throughput.
  • Treating the number as the whole story. The other timings and the platform matter too.

What next

Use this often? Add RAM Latency to your Home screen.

Frequently asked questions

What is a RAM latency calculator?

A tool that converts CAS latency from clock cycles into nanoseconds, so that memory of different speeds and generations can be compared.

How do I convert CAS latency to nanoseconds?

ns = CL × 2,000 ÷ the speed in MT/s. DDR4-3200 CL16 is 10 ns.

Is DDR5-6000 CL30 faster than DDR4-3200 CL16?

Both have a first-word latency of 10 ns, but DDR5-6000 transfers data much faster.

What is the clock of DDR4-3200?

1,600 MHz. DDR memory makes two transfers per clock, so the clock is half of the transfer rate.

What do tRCD and tRP mean?

Row timings in cycles. With no row open, the delay to the first data is tRCD + CL, and with the wrong row open it is tRP + tRCD + CL.

What went wrong?

We only receive the reason you pick, never what you typed into the tool.

Thanks for your feedback.

More Physics & Science Tools