Your Ryzen 5600x seems to be acting strangely.
Your Ryzen 5600x seems to be acting strangely.
I doubt it until several people confirm the same point. Any references or real examples of individuals experiencing this issue? It’s hard to alter settings just by changing the XMP profile or using OS tools to tweak timing. Right now, I’d recommend using 4.5-4.7g static allcore OC at 1.25-1.35V. Don’t go higher because returns drop off and voltage headroom is minimal (max 1.45V). This should push the CPU to run faster and improve performance. Static OC can’t be changed via clock stretching, so if this was causing poor CPU speed, it should resolve it. RAM frequency (3800-4200) and TRFC settings matter a lot—they perform well in tightening. Only adjust the primary timings if needed, except to lower the voltage requirement. Raising TRFC might help distribute load on the IMX, but be cautious with the IMX’s heavy load there. If OC still doesn’t work even at 1.2V vsoc, try loosening the TRFC.
I'm checking online archives from a while ago when CJR was more popular. A few individuals experienced this, and considering how many actually try memory overclocking at higher voltages, it's clear it's not advisable. Don't trust tREFI being managed by the XMP profile—it's handled by the motherboard, and no OS software allows timing changes on AMD RAM except Ryzen Master, which can't adjust that. It's just one of the oddities about overclocking Ryzen memory.
Now I'm genuinely convinced it's not a coincidence. I've also spoken to others who handle higher voltage settings and haven't encountered any issues. Doubts about degradation claims are mostly based on articles that lack credible sources or are written by a single person possibly misinterpreting results (like someone damaging a 45nm C2Q or C2X chip at 1.45V, which might be due to a dirty board or a misleading TLP claim). People testing in the 1.4-1.5V range have reported no degradation either, supporting the idea that these claims aren't universally valid.
Man... What do you understand about X58 and all that degradation chatter? I haven't seen any screenshots from you on any platform claiming these figures. None. No proof or evidence of your personal stories. So far, I'm not convinced and it's tough for me to trust what you say. You're pushing TPU articles again? Let me give you a real-world example of what 3rd place looks like with X58, so you can get my perspective instead of just believing the hype. EDIT: 1024m score.
Oops, I didn't test the new settings before selling both my x5660 and now my p6t deluxe v2. It looks like the 8-hour 2.46V DDR3 Prime95 run wasn't even started, and the photos disappeared except for a weak 2V DDR3 setup that needed 1.55V to function. I mostly kept it at 3900 MHz instead of the higher 4000, since I thought it wouldn’t hurt much for longevity. I’m considering getting a gigabyte X58 soon after seeing an i7 920 at $31, and I plan to try 1.6 and 1.7V VTT (32nm uses about 0.05V less than 45nm). I might switch to an x5650 if it helps cooling performance, though it’s still not ideal. The 4.8GHz clock feels a bit low for third place globally—assuming it’s ambient cooled.
Keep this image in mind, use it when you reach your 920. Compare it to my 930. You crushed that run—I’ll give you a shoutout. Don’t worry about the frequency or all the details. Focus on the 9.095s and nail it. What I’m saying is: you did it, and I’ll reward you with an X58 chip from my stash. Maybe a few more. And some Xeons would be nice too.
Wow, the specs are around 2200 max. I think it’s related to the CPU, even my PS3 couldn’t push past that. Apparently Xeons have terrible IMC, but compared to the already weak X58 i7, going higher makes sense. You mentioned a 2850 seems high for an X58, especially if it’s running in ambient or subzero temps. Based on the timings, it might be a high-end Hi-NR 4Gbit processor. For the X5660, I see only 4.85g booting on 2 cores at 1.7v, which is pretty low. I wonder if I7 chips generally outperform Xeons overall, especially since your X5660 looks like it’s barely running. You said you probably won’t exceed 4000 cores, but scaling seems rough after 3900. At 1.7v with 2 cores off, you’re not pushing it too far. I doubt you’ll beat it, but maybe a 920 would work better. It might need another PS3 if you use it as a reference, and I’m not sure if 3GB RAM is enough. Cooling should be fine, though. So probably a 920 with no other CPUs and all cores enabled?
Anyone can achieve that score, but it doesn't look impressive. Need some tips for epeen? It's mainly high frequency, but triple channel is essential. Turn off HT unless you really need it for that benchmark. 1024m Wprime, all 12 threads on a 980 is intense. A lot of heat. For a Pimod test, reduce to HT off and use 2 cores. Windows XP lite works fine. Push the CPU speed. I'll recommend the lowest water loop if you're aiming to surpass my 5.3ghz X58 setup—better get into LN2 and we can do a proper run together.