DDR4, Alder Lake architecture may experience memory controller issues over time.
DDR4, Alder Lake architecture may experience memory controller issues over time.
I don't have the F4-4000C16D-32GTZR set. I used the F4-3200C15D-32GTZ which I had in stock because I couldn't get XMP to function properly. I didn't realize not all my components were compatible with 3200Mhz—it was my first build. If that RAM sticks, I'll keep using it. I was curious about the other set just in case my current one is faulty (my postcodes suggest it might be RAM or slots).
The difference is clear—this model performs significantly better in TRCF and other timing metrics. The devices I've tested consistently run at 15 MHz at 1.5V, while the 4400MHz unit handles 4000MHz TRCF 13. tCL matters, but TRCF remains the top priority; other sub-timings still boost overall performance even with slightly higher CAS latency. Look for another memory solution that matches these specifications closely, as highlighted in the details.
This indicates a shift in capabilities, as 3200MHz was the maximum supported by early boards. Success often depended on having a capable CPU to run them. Your setup likely wouldn't have worked, and the board probably failed. X99 was a relatively new platform at that point, and it's unusual for it to be failing after just seven years.
I’d check the rating first, 1.4v should work without problems. 1.5v is also acceptable, though if you’re concerned about cooling it might be better to stick with 1.4–1.45.
I purchased a 12900K and the AIO isn't upgrading properly. I'm not sure which cooler to choose yet—I haven't decided on a cooler yet. I still have my NH D-15 with a bracket, but some people think it won't handle the 12900K well. They suggest either a front triple 140 intake or a side triple 120 intake setup. I think the airflow for the RAM should be fine, just need to pick a cooler. My GPU is running hot but not extremely so.
On my DDR2 Extreme Dark system, I need to fully adjust the TRFC to 132 for noticeable performance gains over 1300. Also, I should relax the TRFC initially as a RAM OC step because most setups use higher frequencies with looser TRFC.
I mixed up tRFC and tRCD a lot. Still has an edge with tRCD, it really helps B perform well, clocks pretty accurately, and those results are top-notch.
Some of the sub timings will still be decent, but overall no. Good B die is really good, bad B die is really bad. Not really much in the middle though. Granted, all the kits I've worked with have been solid B die bins (2 G.Skill 3200MHz CL14 kits, one dual rank one single, one kit of Adata XPG 3600MHz CL17-18-18-38 single rank, 1 dual rank kit 4400MHz CL17-18-18-38), but they have all done very respectable subtimings. That said, with how bad I've seen bad B die be, it really should be avoided at all costs. That said, used B die can be pretty cheap (I haven't paid more than $100 for a kit of B die to this date, including that dual rank 4400MHz kit) and as long as you get it in a solid B die bin from someone like G.Skill it'll be very good. If you have to buy new though, Micron Rev B is almost certainly the better buy since the performance outside of a couple benchmarks is very similar and it's a lot easier to work. Plus cheaper new.