Significant variations exist in stress tests.
Significant variations exist in stress tests.
Hello, I'm working on identifying the optimal setting for overclocking my I5 6600K. The issue is that results vary widely across different stress test programs. I've tried: Prime95 small FFT, Prime95 Blend OCCT Small, Extreme, AVX2 OCCT Large, Normal, AIDA64Extreme FPU, AIDA64Extreme CPU, FPU, cache, system memory, and local disks. Temperatures stay below 60°C at 1.35V max, which seems good. However, errors appear during stress tests, but they differ a lot. This wide range makes it hard to decide which software or settings will give a stable sweet spot. For example, 4.3 GHz at 1.28V causes errors in just a few minutes with OCCT Small and Extreme using AVX2, while 4.5 GHz at 1.35V works fine with OCCT Small and Extreme on the same architecture. I don't see errors when running at 4.5 GHz at 1.33V with OCCT Large, Normal, or SSE. Prime95 crashes even at lower frequencies.
I'm considering using AIDA64Extreme or OCCT Large, Normal, SSE for a stable run, especially since gaming won't demand such high loads. I'm not sure if setting the vcore above 1.4 is safe with low temperatures, or if a higher vcore could still risk damage despite stable temps. My current setup uses a fixed VCORE with LLC auto in Gigabyte K170 K3 Gaming motherboard. Please let me know your thoughts!
Are you ready to handle crashes in games? That’s a decision you need to consider yourself. Systems vary in how stressful they are, which is why I often check for the worst-case scenario. High Vcore might harm CPUs even if temps stay low. It's still a balance between speed and durability. A 1.4V setting should extend its life beyond its performance limits, and I wouldn’t push it higher.
Sure, a 6600k works well. Going past 1.5v is acceptable and usually considered lucky for cooling. Above ~1.4v the CPU might start to degrade over time, but staying in the 1.4-1.45 range won’t cause major issues.
I face a challenging situation since I lack the expertise to compare stress test results with actual gameplay demands. Many have expressed concerns that high-stress tools like prime95 and OCCT on extreme conditions are unrealistic, suggesting they shouldn't be relied upon. Conversely, others aim for maximum stability by pushing systems to their limits.
System stability generally means the capacity to execute any program reliably for as long as needed. Prime95 is an extremely demanding benchmark to complete, yet if your system succeeds, it should handle most tasks. Avoiding such tests carries its own risks—just keep in mind that skipping them isn't without consequence. If you're okay with occasional crashes, running your processor at full capacity makes sense, but for peace of mind, ensure Prime95 passes.
You're unsure how much extra load Prime95 adds compared to your games. If the impact is modest—around 5-10%—it makes sense to opt for a stable version like Prime95 Stable. However, if Prime95 is significantly heavier than your applications, a more lightweight option such as Prime95 AIDA64 might suit you better. I’m not focused on overclocking and don’t plan to run resource-intensive 3D or heavy software. What matters most is maintaining stability during games. If you can only reach 4.2 GHz due to Prime95 overhead, while keeping temps around 60°C and avoiding BSOD, AIDA64 with an AVX processor would run smoothly at 4.5 GHz. That seems like the ideal choice.
I switch to OCCT now rather than Prime95 since it halts automatically when temperatures rise or issues appear. I've learned the latest Prime versions can harm the CPU. My configuration includes: Data Set – Small (more weight than Medium or Large), Mode – Extreme (heavier than Normal), Instruction set – AVX (only AVX2 adds extra load, not for games). Is this challenging enough to verify stability?