Question CPU Temp goes up to 90 degrees at startup (linear increase not sudden)


Nov 22, 2015
So, I live in Lebanon and my pc was next to the window which is less than a kilometer away from the explosion that happened in august and there are no buildings between that window and the site of the explosion. I managed to recover my pc, took it apart and cleaned it and rebuilt it and it ran fine. Around a month later, I got a CPU Fan error and when I fixed that, I started getting a CPU Over Temp error which I couldn't fix at the time and ordering parts or thermal paste was extremely hard at the time (international shipping is extremely slow and very expensive without mentioning that thermal paste in Lebanon costs more due to the hyperinflation).

So, now I managed to get thermal paste and replaced the one I had (it was pre-applied, I have a Corsair H100i GTX) and I still get high temperature at startup. When I boot up the computer and go to the BIOS, I can see the temperature rising linearly, when I saw it got to 90, I immediately turned it off. I kept the PC off all night, next day, I turned it on and tried to install and access iCue as fast as possible to check if my cooler is working fine but iCue didn't detect it so I turned off again to avoid damaging the CPU.

Next day, I installed Corsair Link (older version since now it is called iCue) and it detected my cooler but said it's being managed by iCue. So, again, I immediately turned off the PC. I have no clue what to do, I'm avoiding turning on the PC because I can't afford having to replace my CPU AND Cooler (if the issue is my cooler). So, I'd appreciate any advice on either finding out whether the issue is from my cooler, because if it is, then I'll save up and get a new one, but I don't want to spend money on a new one, install it and get the same error again.

I think my socket is well installed, I've tightened the screws in a criss cross pattern and used my phone camera to look if there was a gap between the heatsink and my CPU and it looked like they were in contact.


What you describe sounds a LOT like the PUMP has failed. There are two types of info you can look for.

1. iCUE or LINK should show you the pump speed if you have connected the system correctly. Alternatively WITHOUT using either of those, IF the wire from the PUMP to the mobo is plugged into the CPU_FAN header, boot and go directly into BIOS Setup to where fan headers are configured. Look at the CPU_FAN header. It will tell you the speed of its "fan", which really is the PUMP plugged into that header. If either of these shows that speed to be zero, you probably have a failed pump.

2. BUT there is another possibility that will not be shown as a zero pump speed - a leak that has causes loss of liquid from the cooling loop. This test may tell you about that OR give you the same result if there is NO liquid flow because the pump has failed. Start with your system cold. Open the case so you can reach in and feel the two hoses coming out of the pump, and also at their other ends where they enter the radiator. Start your system and, while it warms up and you can see temperature readings on your screen, feel each hose right where it comes out of the pump body, and again right where it enters the rad. In normal proper operation, at the pump end ONE of the hoses will get a bit warmer than the other, but not really hot. IF, instead, you find that BOTH hose ends get warmer and warmer (until your screen says the CPU temperature is too high and you have to shut down), that indicates NO fluid flow. Meanwhile, at the rad ends of those hoses, normal operation means that again, ONE of them will get warmer than the other. BUT if neither of them warms up, then there is no fluid flow.

If these tests say the pump has failed, OR if it is working but there still is no fluid flow, the only solution is to replace the system. Even if the pump is working, it is extremely difficult to open the tubing, refill the system, and seal up again.

See my additional post below.
Last edited:


BEFORE you go replacing stuff, do this. I know you have checked your connections, but I suggest you do that again.You would hate to spend time and money replacing parts that were not actually faulty.

The Corsair H100i GTX system has three electrical cable connections to be made:
(a) a cable from the pump to the CPU_FAN header with three holes in its connector. It MUST go to that header so that the iCUE software can read the PUMP speed from hat header. The system gets power for the pump and rad fans from there.
(b) two rad fans must plug into connectors on a cable coming out of the pump. This is how iCUE can control their speeds.
(c) a cable goes from a socket on the pump to a mobo USB2 header. This is the link by which the iCUE tool can communicate with the pump and the rest of the H100i system.

First, shut off your system and open the case. For each of those connection points do this. Carefully DISconnect and visually inspect whatever you can see for any obvious dirt or flaws. RE-connect. Repeat several times. Go on to the next connection point. In some cases there may be small amounts of dirt or oxidation on contact points, and this process may "scrub" those surfaces clean to re-establish a good connection. In doing this you also will manage to disturb and re-connect something that just MIGHT have been loose, but not obvious from a visual inspection. When done, look around closely for anything you might have disturbed by mistake, then close up.

If you have not already done that tests above on your system, do this electrical connection thing first just in case it actually fixes the problem. If you HAVE done those other tests already, do this process and then re-try those performance tests from a cold system, in case there is a change. By the time you have done all of this, you will know much more about what may have failed.