I'm considering upgrading from a system with an Intel i7-10700K, Radeon RX 6600 XT, and 32GB of DDR4-3200 CL16 to one with an Intel i9-14900KF, GeForce RTX 4080 Super, and 32GB of DDR5-5600 CL46. The main use is graphics- and shader-intensive gaming, especially heavily modded Minecraft with demanding shaders. DDR5-5600 CL46 has a noticeably higher CAS latency than DDR4-3200 CL16, so would that make it slower in practice, or do the higher transfer rate and bandwidth outweigh the latency difference?
4 Answers
A useful way to compare kits is to calculate CAS latency in nanoseconds: latency is roughly CL multiplied by 2000 and divided by the memory speed. That gives about 10 ns for DDR4-3200 CL16 and about 16.4 ns for DDR5-5600 CL46. DDR5 still offers much more bandwidth, but this particular CL46 kit is relatively loose. If the price difference is small, DDR5-6000 or DDR5-6400 with tighter timings would generally be a better pairing for a high-end system, assuming the motherboard and CPU support it.
The comparison is heavily overshadowed by the rest of the hardware. Going from a 10700K and RX 6600 XT to a 14900KF and RTX 4080 Super will be a massive upgrade regardless of the memory latency. In most graphics-heavy games, the GPU will matter far more, and DDR5-5600 should perform perfectly well. You might see small differences in CPU-limited frame rates or 1% lows, but not enough to make the new system slower overall.
Memory speed and latency affect the CPU differently depending on the workload. Competitive games with very high frame rates may show a measurable difference, while a heavily shader-bound game is more likely to be limited by the GPU.
For normal single-player gaming, the difference between these memory kits is likely to be minor compared with the CPU and GPU upgrade. Very high-refresh esports titles or CPU-limited simulation games can benefit more from faster and tighter memory, but shader-heavy modded Minecraft may be limited by shader complexity, rendering settings, or the GPU instead. The DDR5 system should be substantially faster overall.
DDR5-5600 CL46 has higher throughput and bandwidth, but the CAS latency in nanoseconds is also higher. CL is only one part of total memory latency, though, and it doesn’t tell the whole story about how the system will perform. DDR4-3200 CL16 is about 10 ns for CAS latency, while DDR5-5600 CL46 is about 16.4 ns. Despite that, the newer platform’s overall performance will depend much more on the CPU, memory controller, timings, and workload.
Higher transfer rates don’t automatically erase higher latency, but they do provide more bandwidth. Total real-world performance also depends on secondary timings and the platform, so CAS latency alone shouldn’t be used as the final comparison.