Stability Testing Questions
Stability Testing Questions
Hi guys,
I have a few questions and hope you can assist me.
1. Prime95 26.6 is suggested for thermal testing. Why does this relate to AVX? Is using AVX FPU too much? Should I tune my OC in 26.6 for 90 degrees, and if so, will that cause temperatures to reach 110 degrees due to version differences? That sounds like a concern.
2. A short FFT for 10 minutes should suffice to observe peak temperatures. Am I right about that?
3. What’s the best way to stress test? Should I run a large FFT, Blend, or Custom mode? There are many conflicting suggestions online.
4. How can I confirm Prime95 has completed a cycle?
5. Some recommend running p95, IBT, and a CPU-heavy game together, while others suggest using OCCT. I even found a thread about combining Furmark and IBT.
I’m a bit confused and would appreciate your guidance.
if Prime operates at its standard configuration for around 30 minutes, you remain stable and can gauge your peak temperatures. the longest time I experience is roughly 15 minutes. others push it to hours with higher configurations.
run Furmark separately for GPU evaluation. the same applies.
90°C feels extremely high in my view. if I approach 80°C, I’ll lower the clock. perhaps a more effective CPU cooler would help.
if Prime operates at its standard configuration for around 30 minutes, you remain stable and can gauge your peak temperatures. the longest time I experience is roughly 15 minutes. others push it to hours with higher configurations.
run Furmark separately for GPU evaluation. the same applies.
90°C feels extremely high in my view. if I approach 80°C, I’ll lower the clock. perhaps a more effective CPU cooler could help.
JohnBonhamsGhost explains that running Prime for about 30 minutes indicates stability and familiarity with maximum temperatures. The longest he experiences is around 15 minutes. Others may run it for hours with higher configurations. He suggests testing Furmark separately for GPU evaluation, as the same applies. He notes that 90°C is quite high in his view, and if he approaches 80°C he would consider downclocking. He might try a better CPU cooler. Prime uses Blend mode by default, but Small FFT tends to raise temperatures too much. There are reports of errors after 18 hours, which worries him about data consistency. He asks what the difference is between running separately versus together with Prime. He mentions using an NH-D14 cooler and maintaining core temps at 79, 89, 87, and 78 degrees. He explains he plays games, surfs, and runs virtual machines for testing, so no stressing the CPU at those levels. He also shares his current specs: 4.7 GHz, 1.28 Vcore, and notes the temperatures are stable under these conditions.
run Prime to evaluate your CPU performance, run Furmark to assess GPU capabilities. not doing them simultaneously. if you're using built-in graphics, consider testing both together, though it seems you likely have a dedicated GPU given your gaming focus. what programs rely heavily on FFT and cause concern? my 4790K never exceeded 70°C at 4.6GHz/1.25v when Prime was running at default settings, so you're probably within normal ranges. i've never faced problems with over a year of 3DS Max modeling, video editing, or ultra 1440p gaming—those tasks demand the highest processing power this system can handle. the GPU consistently stays under 66°C during Furmark tests and performs well with these applications.
JohnBonhamsGhost provides additional tips: use Prime to check CPU performance, Furmark for GPU testing, but not simultaneously. If using onboard graphics, consider running both together, though a dedicated GPU seems more likely given your gaming focus. Regarding software concerns, your 4790K never exceeded 70°C at 4.6GHz/1.25v with Prime, so you're likely within normal limits. Over a year of 3DS Max modeling, video editing, and ultra 1440p gaming hasn't caused any issues—GPU stays under 66°C during Furmark tests. I'm using discrete graphics and haven't used FFT extensively; I run small FFT sessions for about 10 minutes to gauge peak temperatures. Since I play at 1080p, I'm unsure about stress testing methods. Many recommend default settings for short durations (12 or 24 hours) and separate Furmark runs for duration estimates.
Demer84 asks about trying default mode for extended periods, wondering if it’s sufficient and whether a separate Furmark is needed to track duration. They clarify they’re okay with shorter sessions (15-30 minutes) and suggest monitoring temperatures during games. They recommend keeping an eye on temps and troubleshooting if they exceed safe limits, noting that anything below 70°C under stress is acceptable for them but others may have stricter thresholds. They also mention searching for safe temperatures for overclocked systems.
JohnBonhamsGhost :
erdem84 :
...shortly you suggest me running default mode for several hours (12 or 24) - I'm not sure how much will be enough - and a separate Furmark for how long exactly or approximately?
no, i stated if they run for 15-30mins with no problem then you should be fine.
why exactly are you so concerned? if your system is running fine and your games have no problem then you have nothing to worry about. just keep an eye on the temps while gaming and if they get out of hand then start to troubleshoot. anything below 70°C at stressed levels, like gaming, is good for me but others allow even higher. do a search here for safe temps with an overclocked 3770K.
I have read the Intel Temperature Guide topic and I decided my max as 80 degrees on load. I am still on the OC process and trying to find my max and be on the safe side as much as possible. 100 MHz difference is not so much difference but if I will be able to hit 5 GHz within reasonable limits why shouldn't I? And BTW what is all the fuss about long hours if 30 mins is enough?
Last Minute Update: I started to run IBT with max for 20 times and the temperature has risen to 93 degrees even though it hasn't completed the first test so I stopped. I'm considering about going down a notch
the higher the temperatures, the shorter the lifespan of the CPU. reaching 90+° a few times might not be too bad, but operating at those levels for extended periods will cause serious damage. just a caution.
if you're using your system for heavy tasks such as complex algorithms or large-scale 3D modeling, it's important to ensure your setup can sustain long runs without issues and losing progress. you might want to test with programs like Prime or Furmark for several hours first. this seems to have moved into the overclocking community without clear justification.
JohnBonhamsGhost shared his thoughts on CPU temperature and longevity. He noted that higher temperatures reduce lifespan, warning that frequent hits above 90°C can cause serious damage. He suggests running stress tests like Prime or FurMark to ensure stability before long-term use. He also mentioned considering more demanding tests for intensive tasks and questioned whether virtualization would affect performance.