No clear upgrade; performance gains are minimal and often offset by stability issues.
No clear upgrade; performance gains are minimal and often offset by stability issues.
I recently increased my RAM speed from 3600 to 3733 with excellent timing. It has remained fully stable during numerous tests lasting 10 to 40 minutes in prime 95, using a DRAM calculator memory tester and Memtest Extreme1 preset. I’m checking if it’s more beneficial to keep at 3733 or aim for 3800 performance and lower latency. My current AIDA64 score is still solid. If you prefer staying at 3733, would you suggest any further timing tweaks or settings adjustments for better results? I’m new to RAM overclocking and began experimenting just a week ago after some research.
It's running at 66MHz. If you're mainly focused on getting the best scores or just exploring, you're welcome. For everyday use, it probably doesn't be important.
The difference between 162fps and 160fps is often just a small change unless you're comparing results or trying to get a higher score. In real life, people usually don't notice the difference. Between 3200 and 3800 frames per second is likely a minor adjustment—maybe 3733 to 3800 feels like extra effort without much gain...
This duration wouldn't represent a true stability assessment. Memory evaluations should last around 24 to 48 hours.
I’d experiment with several configurations. Likely I can extract a bit more performance. Latency range 13-14-14-14-34 (just benchmarking?) Read latency may drop below 14. I set the benchmark at 8 for testing. tREFI and (ns) can improve by 100 at once. Values from 14553 to 14400 give an example of 7796.25 to 7696.25. Tasks like tFAW, tRC etc. are mostly trial-and-error depending on the hardware. Adjust gradually. Record every change. ProcODT works best with lower resistance. Fewer changes between tests help. Note everything. Optimize clock drive strengths too. These settings should stabilize with higher voltage density. B-Die performance peaks around 40°C or less, ideally under 45°C at full load. Keep them cool actively. Daily tweaks using this spec (14-14-14-14-34) at 24°C, 44°C, 12V, 288mA, or lower will work. 1T/2T testing is recommended. It’s not worth overclocking aggressively until you hit around 1.6V with your boards. The current I run is 4000mt/s. How far does this B-Die exploration lead? Or is that a fun thought experiment? Just like chewing on a tootsie pop—how many tries before you find the center? Ultimately, every rig behaves differently. No one will give exact numbers, but this should serve as a solid starting point. Good luck!
You're just starting out and already doing great work. Why not save your 3733 as part of your profile in your bios for a backup, and continue experimenting? Besides the benchmarks, you won't even notice a difference... But that's not the main reason we do this. I encourage you to keep going! Watch how far you can go!
I couldn't go any further down than my current settings except by lowering the TRFC to 250. Every adjustment I make now risks losing benchmark points or causing big score delays. I didn’t find a way to change the tREFI, so I’m unsure what to look for in the BIOS. What does ProcODT do? I’ve never heard of it or read anything about it. I’ve enabled downmode, which probably sets the command rate between 1 and 2, meaning CMD2T is around 1.5T. If TRCDRD is set below 15, it leads to memory errors. My memory stays solid at 71/72 degrees fully stable—does that mean normal operation, or are you asking about micro stutters in an example?
That's the reading after stress testing with testmem5 for an hour continuously. It shouldn't be problematic on idle—it's around 40-41°C—but since it's getting warm outside and the temperature is now 26°C, it's warming up. During my regular tasks it stays near 50-52°C.