Issue with DDR5 Ram
Issue with DDR5 Ram
I just assembled my PC outside the case for the first time and everything looks good. The only issue is with my RAM—CL46 5600. In the BIOS, the EXPO settings are set to 5200MHz, but the profiles show different speeds: one says DDR5 5600MT/s at 46-45-45-90 and another at 42-42-42-105. I’m confused because EXPO profile 1 lists CL46, while profile 2 lists CL42. If EXPO profile 2 is actually running at CL42, it seems contradictory since the RAM can only support CL46. And I’m not sure how to change the EXPO profile if it’s stuck on a slower setting. My motherboard is an MSI MAG B650 Tomahawk Wifi and I’m using Crucial DDR5 Pro CL46 5600.
CL stands for CAS Latency and is expressed in clock cycles, meaning values are only truly comparable at identical clock rates. At 5600 MT/s, CL 46 equals approximately 16.4 nanoseconds, while CL 42 at 5200 MT/s comes to about 16.2 nanoseconds. The gap between them isn't substantial. You'll experience a bit slower transfer speed compared to 5600 MT/s, but latency will be marginally lower. This is expected—higher speeds usually mean higher latency. It's a balance you can adjust. Overclocking RAM often requires trade-offs: higher speeds may demand increased latency and voltage for stability, while lowering clock speeds can reduce latency but at the cost of speed.
RAM may operate using any configuration available. Claimed rates and EXPO presets are verified by the maker and should remain consistent. The mainboard uses standard JEDEC rates by default. The listed RAM performance is an overclock, activating either profile when chosen. Consider updating the BIOS to potentially enable the EXPO setting.
It's fine in the sense that nothing is defective. As @WoodenMarker said, a BIOS update might get the second EXPO profile working as well. 5600 MT/s CL 46 would give you ~7.7% higher memory bandwidth at the cost of ~1.2% higher latency compared to 5200 MT/s CL 42. You could test with a few games if there's any measurable difference, but it'll likely be negligible. Synthetic memory benchmarks might show some difference.