Choose between the I7 14700K and the I9 14900K for your build.
Choose between the I7 14700K and the I9 14900K for your build.
We aimed for better latency with DDR 1. While latency matters, OOE CPUs are less critical. When the on-die scheduler detects a cache miss, it will react quickly by handling other tasks. Latency only measures one aspect—how fast a 4Mb page reaches L3—while actual performance depends on factors like cas latency and speed. For example, with DDR5 at 3000Mhz/s, latency plus 4/6000 is close to the real time, whereas DDR4 at 1600Mhz/s adds another factor. If cas delays match, DDR5 can still hit around 670 microseconds for this job instead of a much longer time. (Adjust for bit width: 1 bit per channel or 128-bit per dual channel as needed)
They match exactly, as confirmed by the consistent cas latency. 6000MT CL30 equals 3200MT CL16 in this regard.
Latency only shows how long it takes for a page to reach cache; it doesn’t indicate how quickly the CPU processes a request. For example, a 4Mb page on DDR5 responds in microseconds rather than milliseconds. CPUs prefer latency values because they often handle 4KB pages for general tasks, but large data like gaming can benefit from 4MiB pages when bandwidth is more important than speed. It doesn’t matter if latency is just 10 nanoseconds versus 16 nanoseconds—when a 4MiB page takes microseconds over a network, the delay becomes noticeable. At 6000Mbps, a 4MiB transfer would still take about 10 microseconds. A 4KiB page would be around 100 nanoseconds, and latency really matters because it affects performance as soon as it reaches 110 nanoseconds.
You were inquiring about how DDR4 affects performance. Being on DDR4 restricts how much you can improve your CPU. If a Raptor Lake upgrade boosts speed by 15-20%, but then Raptor Lake suffers a 10% drop, the overall gain drops to just 5-10%. This shows that DDR4 didn’t significantly hinder AlderLake’s performance, so it wasn’t a major issue. I’m not suggesting it’s worth spending extra on the board and RAM right now, since you’re already paying double for upgrades. Just keep this in mind if you decide to upgrade.
The 15-20% upgrade relates to DDR4 and DDR5 technologies. If it involves DDR5, it won’t limit the performance gain; otherwise, you’re missing out on additional improvements. Your confusion seems to stem from wording about using DDR4 versus DDR5, not switching between them.
I own a 13600k paired with a 4070ti super, it’s a great setup too. The same cooler works well, though I had to lower my CPU voltage and push it to 5.3P, 4.2E4, 4.7R at around 1.18V for smooth gaming like 60C and 85C R23. If you go stock out the box, a 100°C will be tough, and downclocking might help. A 14700k could struggle even with these tweaks.