Prime95 trouble
Prime95 trouble
Hey everyone, I'm facing some issues with Prime95. I'm aiming for a 3770k Prime stable at 4.5GHz with 1.26V. My RAM is 2000MHz, and I've been using 10-11-28 (2T) with 1.565V (it's a 2x8GB 2100MHz kit XMP). The temperatures on HWInfo64 are always under 80°C on any sensor. I ran the program for up to two hours without any crashes, even while doing other tasks. Later, I tried Firestrike, Heaven, and some games—no problems there either. After a few minutes, the Prime95 process ended cleanly with no BSOD or errors. I restarted it, and it ran for over 40 minutes without issues. I experimented with different settings: lower voltages, adjusted timings, higher voltages, looser delays, added PLL voltage, etc., but nothing worked. I'm using an ASUS P8Z77-M with the latest BIOS. I'm running out of ideas. I read somewhere that P95 might have problems when used with a different storage device than the OS drive—so I uninstalled and reinstalled it to Desktop, even adding the ZIP file. Someone also suggested enabling Spread Spectrum in BIOS to reduce interference. I haven't tried that yet, but BIOS recommends keeping it off during overclocking. Should I just start the overclock again?
Use Asus Realbench to test performance. P95 often feels like a weak spot now. https://rog.asus.com/rog-pro/realbench-v2-leaderboard/
It has been operating for 5 hours and 45 minutes without problems, though temperatures remain 3–4–5°C below the P95 level. Can you say it’s stable now, or should I think P95 stays more demanding?
P95 demands more power and isn't truly reliable. Running it strains the system, causing crashes and performance drops. I empathize with your concerns—there are times when stability feels uncertain. Despite testing extensively across multiple apps, I've never encountered any problems. My experience shows my setup handles demanding tasks smoothly. P95 doesn't represent a genuine use case for me.
It's understandable you're checking the system's stability. I didn't expect P95 to behave as described—my goal was just to confirm it wouldn't fail unexpectedly for someone unfamiliar with it. Still, I'm hoping it's just a matter of the calculations being challenging rather than a bug.
It depends on how the program executes and the specific AVX instruction demands. I've never faced problems with CPUs older than 6700k; newer Intel chips generally avoid P95 well, but the 3770k might be struggling for some reason. My 4770k and 6700k handled P95 without issues, while my 8700k and 9900k both rejected it. On my 9900k I run at around 80°C in AIDA FPU @5 GHz—immediately jumps to 99°C, throttles, and throws errors in P95.
It might be caused by excessive voltage requirements or too many amps. I’m unsure for certain; I’d anticipate passing P95. However, if you can run auda64 for 24 hours using all CPU and RAM settings (default) and 24 hours of AIDA FPU only, plus the complete 8 hours on ASUS Real Bench, I’d feel confident about the CPU performance.