Consider upgrading to 4000MHz RAM if your Ryzen supports it, especially with a 2000MHz clock speed.
Consider upgrading to 4000MHz RAM if your Ryzen supports it, especially with a 2000MHz clock speed.
I installed 4000mhz on the Ryzen and it’s still running without issues, though it hasn’t crashed yet. I haven’t adjusted the timings or done any fine tuning yet. Should I give it another try to see if I can improve stability, or is it better to skip it? I’ve set the clock speed to 3200 and increased each timing I could, just to get it up. Just checking if tweaking timings would help or if it’s pointless.
It’s not unusual to encounter chips running at 2000MHz FCLK, particularly on single CCDs such as the 5600X, though this remains relatively rare. A simple method to verify stability is to perform a benchmark that heavily relies on FCLK and gradually increase it until performance drops noticeably. Unstable FCLK often manifests in memory stress tests and certain error types, most frequently WHEA errors, but it typically appears first as significant slowdowns. For example, the 5900X with a static 4.55GHz and a tuned dual rank B die rated at 4400C17 would take around 25 seconds in Y cruncher 1B, but reaching 1933MHz would extend that time to nearly five minutes. As long as performance remains steady, it’s worth trying to maintain the higher frequency; otherwise, it may be a dead end. Some voltage adjustments can help achieve better scaling in certain workloads, though they’re not reliable for memory stress testing and are mainly useful during CPU benchmarking. Also, note that most processors can boot at 2000MHz FCLK, as my 5900X supports 2066MHz easily, but achieving that speed doesn’t guarantee stability at those rates.
Consider another test method. The current tool crashes the system, so try a different setup. Also clarify whether you mean VDDG, VDDP, and SOC voltages—low VDDG (.965V) led to severe stuttering at 2000 MHz clock.
Checking Y cruncher performance at 3800MT/s RAM is feasible; running at 4000MT/s is typical. The 2000MHz FCLK instability could cause crashes if ECC fails, making it very unreliable. If it stalls at 3800MT/s, lowering the core clock to around 4.6 should help. Super Pi might also reveal this issue, though Y cruncher usually shows problems sooner. The 1.8V rail is noted as relevant.
The issue seems linked to my aggressive CPU overclocking. I don’t lower the frequency because most real-world tasks don’t match the intense load y-cruncher creates, so stability is more important than raw speed. It’s still quite reliable—Prime95 runs smoothly. I can adjust it down for this test but will raise it afterward. Regarding your questions: the setting you mentioned as “cpu vdd 1.8 voltage” matches what you found in your BIOS. Safe values typically range between 1.7 to 1.9V, depending on your motherboard model.
2000 fclk remains relatively uncommon even for single CCD chips; usually I stop at 1867 fclk for regular use. Fclk issues can appear differently than typical memory problems—potentially causing major performance drops due to hardware safe mode, and Zen 3 relies less on raw memory speed compared to earlier versions. Curious to know if those dimmed outputs are related? If so, a 3800 cl14 profile would be ideal and could easily fit high-end GPUs. Just some thoughts from someone who works extensively with Zen chips. The initial reply was very helpful, offering insight into the challenges of pushing Zen 3 chips and confirming that you're unlikely to break any records without extreme conditions.
Yeah, that's what I was implying you do. Pretty sure that's the voltage you're looking for, sounds about right at least. It's been a while since I've had my Taichi so can't double check, but that does sound how ASRock would call that voltage. I'd want to keep it below 2V for long term use, though at least on my 5900X FCLK kept scaling in something like 3DMark CPU tests up until my X570 Master maxed out at 2.35V IIRC (again, it's been a while since I've benched Ryzen 5000 and not daily'd it), but again, it doesn't really help in memory stress tests, so even with that voltage at insanely high levels it still was not remotely stable at 1933MHz FCLK in memory tests with memory settings that should have been 100% stable. You can try punching in 2V and seeing what happens, if you're just barely unstable it might help, but don't get your hopes up for that.
Focus on SOC and VDDGs. If you can't use y-cruncher, it's best to halt. Stress in games isn't the issue—you're fine handling an unstable OC just to prove you can hit a target clock.
My processor handles prime95 at 4.8 still, which is a bit challenging but manageable. It seems it's more demanding than cinebench before, so I think it's acceptable. For now, I'll keep using the 3800 model and focus on improving my timing. The 4000 might actually run slower due to unstable clock frequency.