Key takeaways
- Comparing CL across different speeds. Always convert to nanoseconds first.
- Running one stick. A single module runs single-channel, which usually costs more performance than any CL step. Use two matched modules.
- Forgetting to enable the profile. Memory often runs at a slow default speed until the XMP or EXPO profile is switched on in the BIOS. The menu name varies by motherboard maker, so check the board manual.
- Buying beyond what the platform supports. A very fast kit may fall back to a lower speed on some CPUs and boards. Look at the motherboard maker’s memory support list.
- Mixing kits. Two different kits may not run at the rated speed together, so buy a single two-module kit.
For gaming, the lowest actual latency in nanoseconds wins, not the lowest CL number on its own, and within the same memory speed CL16 is faster than CL18, which is faster than CL20. The catch is that the gap is small in most games, so a faster data rate, enough capacity and a stable dual-channel setup usually matter more than chasing a tighter CL label.
Quick Answer: What CL Actually Means
CL is CAS latency, the number of memory clock cycles the module waits between receiving a read command and starting to deliver data. Because a cycle gets shorter as memory speed rises, a CL value only means something when you know the data rate next to it. A reference article on CAS latency makes the same point: when comparing different clock speeds, latencies have to be translated into absolute time, because a higher CL number can still take less time if the clock is faster.
The conversion is simple. First-word latency in nanoseconds equals the CL multiplied by 2000, divided by the data rate in MT/s. The same reference gives DDR4-3200 CL16 as 10 ns and DDR4-4000 CL16 as 8 ns, which matches that formula.
CL16 vs CL18 vs CL20 at the Same Speed
The table below applies the formula to common DDR4 speeds. These are calculated figures, not benchmark results.
| Data rate | CL16 | CL18 | CL20 |
|---|---|---|---|
| DDR4-3200 | 10.00 ns | 11.25 ns | 12.50 ns |
| DDR4-3600 | 8.89 ns | 10.00 ns | 11.11 ns |
| DDR4-4000 | 8.00 ns | 9.00 ns | 10.00 ns |
Two things stand out. First, at one speed each step of two CL points adds roughly 1 to 2.5 ns, depending on the data rate. Second, a kit with a higher CL can beat a lower-CL kit running slower: DDR4-3600 CL18 and DDR4-3200 CL16 both land at 10 ns for first-word latency, yet the 3600 kit has more bandwidth. DDR4-4000 CL20 is also equal to DDR4-3200 CL16 on latency alone.
Why the Difference Is Often Small in Games
Memory latency is one part of how fast a CPU gets data. Cache hierarchy, memory controller design, how many ranks are installed and how the game itself accesses data all change the result. Reviewers who compare memory kits commonly report that games running at high frame rates with a fast GPU show the biggest gaps, while GPU-limited scenes at high resolution show little change. Treat any claim of a fixed percentage gain with caution, because it depends on the CPU, the GPU and the title.
Processors with a large amount of cache, such as the X3D Ryzen chips, are generally described by reviewers as less sensitive to memory tuning than chips without extra cache, since more data stays on the processor. If you own one of those, the case for paying extra for tight timings is weaker.
What About DDR5?
The CL16, CL18 and CL20 labels mostly belong to DDR4. DDR5 kits typically list much higher CL values, because the clock is faster, so the number looks worse while the real time can be similar. Applying the same formula, DDR5-6000 CL30 works out to 10 ns, CL36 to 12 ns and CL40 to about 13.3 ns. If you are on a platform that only takes DDR5, compare kits by the nanosecond figure and the data rate, not by matching the CL16-to-CL20 range from the DDR4 era. A CL16 DDR5 kit is not a realistic target to look for.
Which Should You Pick?
| Situation | Better choice | Reason |
|---|---|---|
| Building on DDR4 with a mainstream CPU | DDR4-3600 CL16 or CL18 | Good balance of data rate and latency; widely supported |
| Kits cost the same, same speed | Lower CL | Same bandwidth, lower latency |
| CL20 kit is much cheaper at a higher speed | Compare the ns result | Higher speed can cancel the CL penalty |
| Older DDR4 system, GPU-limited at high resolution | Any stable kit, 16 GB or more | Capacity and dual-channel matter more |
| Tight budget | CL18 at 3200 or 3600 | Small gap versus CL16 in most scenarios |
Common Mistakes
- Comparing CL across different speeds. Always convert to nanoseconds first.
- Running one stick. A single module runs single-channel, which usually costs more performance than any CL step. Use two matched modules.
- Forgetting to enable the profile. Memory often runs at a slow default speed until the XMP or EXPO profile is switched on in the BIOS. The menu name varies by motherboard maker, so check the board manual.
- Buying beyond what the platform supports. A very fast kit may fall back to a lower speed on some CPUs and boards. Look at the motherboard maker’s memory support list.
- Mixing kits. Two different kits may not run at the rated speed together, so buy a single two-module kit.
Stability and Safety Notes
Enabling XMP or EXPO is technically an overclock of the memory and, depending on the maker, may fall outside the base processor specification. Manual timing changes can cause crashes or corrupted data if settings are unstable, and some makers state that overclocking can affect warranty coverage. If your system becomes unstable after a change, return to default settings and consult the motherboard manual or maker support.
Alternatives to Chasing Lower CL
- Add capacity. Moving from 16 GB to 32 GB helps more in modern titles that fill memory than trimming 1 ns of latency.
- Upgrade the GPU. When a game is limited by the graphics card, memory timing changes rarely show up.
- Enable dual-channel. Check both modules sit in the slots the motherboard manual names for two-module use.
- Move platforms. If you are building new, DDR5 platforms offer more headroom, though the best choice depends on the CPU you want.
Frequently Asked Questions
Is CL16 always better than CL18?
Only when the data rate is the same. At the same speed, CL16 has lower first-word latency than CL18. If the CL18 kit runs at a clearly higher speed, the real latency can be equal or better, so convert both to nanoseconds.
Is CL20 bad for gaming?
Not by itself. CL20 at a high data rate can match the real latency of a slower CL16 kit, and many games show only small differences between memory kits. It becomes a weak choice when it is paired with a low speed such as DDR4-2666 and a fast graphics card.
Does lower CL matter more than higher MHz?
Neither dominates on its own. Data rate raises bandwidth and also shrinks cycle time, while CL sets how many cycles of waiting there are. Reviewers generally judge kits by the combination, using the nanosecond figure plus the speed.
Should I buy DDR4 CL16 or move to DDR5?
That depends on the motherboard and CPU, because DDR4 and DDR5 modules are not interchangeable. If you already own a DDR4 board, a good DDR4 kit is the practical upgrade. For a new build, check which memory type your chosen platform supports.
Will memory timings fix low frame rates?
Rarely. Low frame rates usually come from the GPU, the CPU, a single-channel setup or too little memory. Look at those first, and treat timing changes as a small refinement.
FAQ
What About DDR5?
Is CL16 always better than CL18?
Is CL20 bad for gaming?
Does lower CL matter more than higher MHz?
Should I buy DDR4 CL16 or move to DDR5?
Will memory timings fix low frame rates?
More from Amazon
Programs that pair well with the guide above. Offers and eligibility are set by Amazon and may change.
Free shipping on eligible items, Prime Video and more. Eligible new members can start with a free trial.
Start a Prime free trialAd-free music streaming. Eligible new subscribers can start with a free trial.
Try Music Unlimited freeA Prime membership for ages 18-24 with a trial period. Eligibility is set by Amazon.
See Prime for Young AdultsAudiobooks and podcasts to listen to on the go. Eligible new members can start with a free trial.
Try Audible freeAs an Amazon Associate we earn from qualifying purchases.