I'd like your input on an issue that me and a friend have with our recently built Core i7 860 based machines. The issue is that we're seeing very high CPU temperatures on both our builds, whether we're using the stock cooler or Scythe Ninja 2/Thermalright Ultra-120 Extreme. We use Prime95 to generate full load on all cores and with the stock cooler and at stock frequency (2.93GHz - Turbo Mode on), we both get temps of ~85C. I have the Ultra-120 Extreme and when I switch to that, the temp drops to around 62C. The fan I'm using spins at 1050RPM. My friend has a Ninja 2 and a fan running on a similar speed and he gets around 65C. We've used a few different programs for measuring the temps, but they seem to give similar readings. Right now, we're using Everest.
Now, although these temperatures still seem a little high considering the good coolers, they would be okay for stock operation. However, both me and my friend are overclockers and this is where things start getting nasty. The temperature rises rapidly when we start upping the frequency. For example, at 3.5GHz and 1.24V, my CPU full loads at ~81C. I've tried connecting a slightly faster fan (1550RPM), but it had only negligible impact on the temperature (maybe 2C). We have both checked the contact to the heatspreader and our coolers do seem to make good contact.
Interestingly, the coolers do not get as hot as they should be if the CPU really output enough heat to reach 80+C (and assuming good contact). After one hour of Prime95, the base of my cooler was hot, but not that hot. If I kept my finger on it for a little while, it started to hurt a little, but it was nowhere near as hot as I'd expect with a CPU running at over 80C. My friend has noticed the exact same thing on his Ninja.
The only reasons for this that we can think of is that the CPU die has bad contact with the heatspreader, or that the temperature sensors don't really work as intended. We've read around on the net and there seem to be a few people with similar temps, but there also seem to be a lot of people that can reach 4GHz on air. I definitely wouldn't dare going over 3.6GHz on this setup... So, my question is if you've seen any similar behavior or heard anything along these lines before?
Perhaps I should e-mail Intel directly about this issue, but I don't give much for consumer tech support from big companies. Plus, they're probably not that eager to help when they hear that the CPU actually does work as intended. I guess I could claim that the CPU actually gets 10-15C hotter than the maximum allowed temp (which is ~73C according to Intel), but they'd probably just say that it's fine as long as it works...
EDIT: See sig for full system info.
Now, although these temperatures still seem a little high considering the good coolers, they would be okay for stock operation. However, both me and my friend are overclockers and this is where things start getting nasty. The temperature rises rapidly when we start upping the frequency. For example, at 3.5GHz and 1.24V, my CPU full loads at ~81C. I've tried connecting a slightly faster fan (1550RPM), but it had only negligible impact on the temperature (maybe 2C). We have both checked the contact to the heatspreader and our coolers do seem to make good contact.
Interestingly, the coolers do not get as hot as they should be if the CPU really output enough heat to reach 80+C (and assuming good contact). After one hour of Prime95, the base of my cooler was hot, but not that hot. If I kept my finger on it for a little while, it started to hurt a little, but it was nowhere near as hot as I'd expect with a CPU running at over 80C. My friend has noticed the exact same thing on his Ninja.
The only reasons for this that we can think of is that the CPU die has bad contact with the heatspreader, or that the temperature sensors don't really work as intended. We've read around on the net and there seem to be a few people with similar temps, but there also seem to be a lot of people that can reach 4GHz on air. I definitely wouldn't dare going over 3.6GHz on this setup... So, my question is if you've seen any similar behavior or heard anything along these lines before?
Perhaps I should e-mail Intel directly about this issue, but I don't give much for consumer tech support from big companies. Plus, they're probably not that eager to help when they hear that the CPU actually does work as intended. I guess I could claim that the CPU actually gets 10-15C hotter than the maximum allowed temp (which is ~73C according to Intel), but they'd probably just say that it's fine as long as it works...
EDIT: See sig for full system info.
Last edited: