10600K Thermals – any issues?
10600K Thermals – any issues?
Thanks a lot everyone, particularly @Mister Woof. I was nervous about getting a faulty chip or thinking the AIO was failing completely. The temperature differences between the 6700k and 10600k were sufficient to raise my concerns. I think I’ll tweak the cooling setup—maybe add better fans—to bring things back to normal levels. As my wife mentioned, it should sound more stable, not like the PC is about to overheat.
Most games won't handle more than 8 threads or cores. The program consumes full memory, CPU core usage, and cache—100% during heavy stress tests, depending on requirements. Edit: I think OCCT supports AVX AMD versions, AVX, AVX2, or without certain instruction sets. I usually run it for Linpack and big data sets. 12GB installed for 16GB. I do this to prevent windblows from forcing the test into the page file. Edited March 9, 2021 by ShrimpBrime Additional Comment
It seems some users prefer older versions of P95 or test environments without AVX to avoid the extra performance boost from newer features. They appreciate the realism and balanced stress on hardware, even though the thermal output is much higher than typical games.
I only stress test at realistic clocks. It's a beating on the hardware. There is such a thing as too much testing imo. For example.... I've watched FurMark completely KILL many many GPUs through the years. dozens and dozens. I stopped using it many many years ago. Either it passes the benchmark or it doesn't adjust accordingly. (But I'm benching, not looking at daily clocks) Daily clocks, I leave my cards stock. A lot of times the CPU as well. Main rig is treated like a baby. I need the redundancy more than a few % additional in gains.
Many people find it odd how testing methods vary, like comparing them to car reliability tests. Sure, manufacturers must check things during development and quality control, but once production starts, those checks stop. This metric changes based on the use case. A light-duty truck’s acceptable MTBF won’t match a heavy-duty one, and forcing a light-duty vehicle to meet heavy-duty standards isn’t practical—it just stresses parts beyond their intended limits. It’s similar to overclocking; people might tweak things for quick results but ignore long-term durability. I see some folks doing minor modifications or tests, but no one runs engines at full power to verify performance. There are also online claims about needing extreme tests, like passing 24 hours of P95, which seems unrealistic. I wouldn’t subject my rifle to thousands of shots just to check the barrel, because that would damage it instead of ensuring quality.
I’ve received an update about the recent improvements. To tackle the noise problem, I made some significant hardware changes. Because my AIO was vented at the top, I switched to a front-mounted unit. Given that my Corsair 540 supports a 360 radiator while my H100i v2 only had a 240, I upgraded together. The new AIO is an EVGA CLC 360, which was available for under $100 on Amazon. My previous setup used a push-pull fan sandwich with four fans, whereas the EVGA model has three. While not identical, they should perform similarly. The results were impressive: peak temperatures dropped from 87°C to 70°C under full load, averaging around 82°C down to 61°C, and low temps fell from 74°C to 53°C. My H100i v2 had been running for nearly four years—so it might have been nearing the end. Perhaps it ran low on fluid or something was failing, and moving it to a new board caused the issue. I didn’t notice any warning signs beforehand; it just quietly stopped working without any noise.