7700k 1.25v 4.7GHZ temp
7700k 1.25v 4.7GHZ temp
Hello everyone, I have a query regarding the temperature of my 7700k.
Here are my specifications:
MSI Z270 Gaming Pro Carbon
i7 7700k OCed to 4.7GHZ
Corsair H100i v1 cooler
In BIOS, I adjusted Vcore at 1.18v, adding +0.05v for all other voltages to the default.
In HWinfo while under stress, Vcore reads 1.235v. Everything remains stable—passed an hour with OCCT large package, ten turns of IBT at max settings, one hour with RealBench, and one hour with Prime95 without any issues.
Before removing the CPU, idle temps were around 46-50°C, full load reached about 85°C with IBT in core 1, while other cores peaked at 88°C with IBT.
I also removed my CPU, using Noctua NT-H1 between IHS and die; core 1 dropped to 78-81°C, other cores around 85°C with IBT.
After switching to Liquid Pro, idle temps are now 43-45°C, full load hits 71-72°C in core 1, but core 4 is 81-82°C—this variation seems normal?
I ran RealBench: core 1 at 66-70°C, core 4 at 4-6°C hotter. In idle, the difference stays between 1-3°C, depending on conditions.
The room temperature is about 22°C inside my bedroom, and my case is NZXT Mid Tower S340.
Even after improvements from cooling changes, I still feel the temps aren’t ideal. Some users manage similar temperatures with higher Vcore and better coolers.
Do you think my current readings are acceptable?
I wouldn't be satisfied with those outcomes either. Even at 71-72 at that voltage, it's quite high for liquid metal. I can achieve 5.0GHz @ 1.4V with NH-D15S and reach up to 73-76C while running Prime95 for 30 minutes. At 1.35V, the max is 67C. (Thermal Grizzly Conductonaut)
My temperatures decreased by about 20°C on average.
I believe I may have used a bit too much when applying the LM. How much were you using? Did you cover the whole die? Were there any areas under the IHS?
volkgren :
I wouldn't be satisfied with those outcomes either. Even at 71-72 at that voltage, it's quite high for liquid metal. I can run 5.0GHz @ 1.4V with NH-D15S and reach up to 73-76C using Prime95 for 30 minutes. At 1.35V, the max is 67C. My temperatures dropped about 20°C on average. I thought I used too much when applying the LM. How much did you use? Did you cover the whole die? Did you place some underneath the IHS? I just applied a little bit, spread the LM across the die with cotton buds, but it didn't look like liquid at all. After that, I placed a small drop on the IHS and didn't re-glue it; instead, I used a clamp on the mainboard to keep it steady on the PCB. Do you think I should apply a bit more of LM?
Did you remove all the silicon holding the IHS in place to ensure proper contact between the IHS and the die? You mentioned you still have the clamp, how does that function? I’m not sure about applying more; I assumed a small amount would suffice after trying it without knowing.
volkgren :
Did you remove all the silicon holding the IHS so it ensures good contact with the die? Do you still have the clamp attached? How does that process work?
Yes, definitely—I cleaned every single bit.
No, I just used the CPU socket on the mainboard, not the clamp specifically.
😀
I didn’t glue the IHS to the PCB.
Well, I'm not quite an expert on delidding since I've only done it once. It seems like you handled everything correctly. The problem might be unclear unless there isn't enough LM. :/
I'll look for the members who assisted me earlier and direct them to this thread.
It seems you're dealing with a minor LM issue. If you've taken out the black silicone from both the substrate and IHS, then the gap between the IHS and the die is as tight as possible—unless you consider it excessive (overkill move). For me, it would be one of two things:
1. Too little LM
2. CPU cooler installed incorrectly.
My idle temperatures are about 2-3 degrees above ambient. At full load at 5.1Ghz with a Vcore of 1.328, I stay around 50-55°C even when the ambient is 20-23°C Celsius.
Hello everyone, thank you for your support.
I just disassembled it, added another tiny drop on the die and the IHS, and tried to spread LM more evenly using a wooden toothpick instead of cotton buds. The surface looks better now, but I still want some "fluidity" and a bit of reflection rather than just leaving scratch marks.
I also tidied up the TIM on cooler, spread them out by hand, and tightened the screws by hand—this time I didn’t use a screwdriver to prevent over-tightening.
Results so far:
- Idle temperature dropped from 43–46°C to stable around 39–40°C.
- Full load achieved in 5 minutes with IBT and OCCT Linpack, reaching about 68–70°C.
- Full load with Realbench took around 70°C, though I think the higher value is due to RealBench stressing the GPU more, while my setup is smaller and ambient temperature rose, raising CPU temps.
- Gaming temperature stayed between 55–60°C.
- The gap between core 1 and core 4 is only 3–4°C, which is a positive sign.
Do you think this performance meets expectations? Should I run another check to fine-tune anything else?
Also, a 4.7GHZ at 1.235v seems decent, but I’m curious—was the small FFT of Prime95 showing a much higher temp (76°C)? I thought SmallFFT was based on Linpack and why was it giving such a high reading?
Hello everyone, thank you all for your support.
I’ve started disassembling and tried placing a small drop on the die, using an IHS, and spreading LM more evenly with a wooden toothpick instead of cotton buds. The surface looks better, but I still maintain some liquidity and a mirror effect rather than just scratching the surface.
I also tidied up the TIM on cooler, spread them out manually, and tightened the screws by hand—this time I avoided using a screwdriver to prevent over-tightening.
Results so far:
- Idle temperature dropped from 43–46°C to stable around 39–40°C.
- Full load achieved in 5 minutes with IBT and OCCT Linpack, reaching about 68–70°C.
- Reabench reached around 70°C, but likely due to RealBench stressing the GPU; my setup is quieter and ambient temperature rise caused CPU temps to climb.
- Gaming temperature stayed between 55–60°C.
- The gap between core 1 and core 4 is only 3–4°C, which is a positive sign.
Do you think these numbers are satisfactory? Should I recheck anything further for improvement?
Also, regarding your question about performance: 4.7GHZ at 1.235v seems decent, but I’m curious—was the Prime95 FFT showing such high readings? Was it based on Linpack, and why did SmallFFT report a higher temperature? Could you tell me which version of Prime95 you’re using?