You can try adjusting BIOS settings and memory timings, but be cautious—don’t skip thorough stability checks.
You can try adjusting BIOS settings and memory timings, but be cautious—don’t skip thorough stability checks.
I'm looking for ways to boost performance without extensive stability checks. I've heard tRFC can significantly affect the AMD platform, but I'm hesitant to experiment alone. Centering on zentimings seems appealing, yet I don't want to overhaul timings or risk damaging my system by raising voltage. Do you have suggestions or adjustments I can make? I'm aiming for lower latency and better single-threaded speed, or perhaps higher FPS in games if that's more feasible. Specs: RTX 3080, Supremacy X MSI X570, unified memory 5900X, DDR4-3600, 32GB RAM, 3600MHz clock, G.Skill 32GB, H100 chipset, 850W power plan.
^^^ Confirmed, used DOCP on most Ryzen setups—it's essentially the same process, just enable it and you're good unless you run into crashes. Rare issues can pop up with certain CPU/mobo pairs and XMP settings. Manually tweaking RAM without proper stability checks isn't practical unless you're comfortable with unpredictable BSODs and constant adjustments. That's why I've skipped manual RAM tuning; my components aren't stable enough to justify the testing time.
Your tRFC appears elevated. Would lowering it slightly—perhaps by about 100 units—be beneficial? Some suggest B-die can manage up to 350, so reducing it by 100 might bring it closer to 450.
If you're comfortable with the chance of random BSODs, then proceed. The risk isn't too big, though they can still occur—if it makes sense to continue.
You can enter any speeds, timings, or voltages manually, but stability might not be guaranteed. The only sure way to confirm performance is through stability testing, which is part of the OC process. If you want faster speeds without going through the full DOCP method, use median values from Zentimings and run a 5-minute test. If it works, proceed with maximum speed; if not, repeat the test. Ultimately, stability testing is essential since RAM sticks vary, and differences in batches can significantly impact results. Trial and error are unavoidable.
1.5v works perfectly. Enable XMP. Manually set the frequency to 3733MHz and you should be fine without heavy testing. Stability preferences vary—perform tests that fit your requirements and use the system as usual. Issues will likely surface only when least expected. Just how these things function lol. Even after continuous operation for a full day, sometimes a restart is needed... just like your phone XD.
I've checked the presets from "memory try it" on my MSI motherboard, but the 3733 didn't work and triggered errors. I'm now using everything as configured, and it seems to be functioning properly so far. Are there any suggestions for testing? I'm not sure if the DRAM calculator or its tester tool is suitable for accurate memory checks.