[SOLVED] Bad thermal paste, or bad cooling?

[tilophase]

Dell Inspiron 15 5000

CPU: i7-8550U

GPU: Radeon 530

My laptop's cooling solutions is overall quite terrible, max CPU temps of ~99°C, temperature immediately shoots up when hit with a load, and constant thermal throttling when under load. Perplexingly, highest I've seen GPU go up to is ~80-90°C, which is most definitely not great, but significantly lower than CPU.

I decided to try fix this by repasting my heatsink. This may have been my mistake, I didn't feel like spending $20 for a tiny tube of thermal paste, so I spent $9 on a tube of Deepcool Z5, seeing some reviews saying that it was cheap, but gets the job done.

Immediately after repasting, thermals for the CPU was quite good. The temperature was not immediately jumping up when under load, it was a curve now, the temperature increased over a decent period of time. I used the benchmark inside XTU, and got a maximum temperature of only 83°C. I ran two more tests over the course of the following hour, and got scores and maximum clocks in about the same range, but the temperature went up to 91°C and 97°C in the second and third runs respectively. I was feeling kind of worried at this point.

It's been about a month now, and when I run the benchmark again, the score is about 30 marks lower, and the maximum temperature is 99°C. The temperature is now back to immediately jumping up and thermal throttling.

Did I just buy terrible thermal paste, and could this problem be potentially fixed if I spent some extra money on something better like Arctic MX-4? Or does this seem like a different issue?

 

[Eximo]

If your first test was from a cold boot, it wasn't a realistic test. Should have waited for the system to reach temperature then test.

Sounds like your multiple bench runs more closely simulated what a real system is like. Takes a while to cool down those heat sinks after they get warm, they are designed for those short bursts that the majority of users experience, not continuous.

I've not heard of that particular paste. Deepcool is somewhat reputable.

Arctic MX-4is a very good general paste for the money. Not the best, but far from the worst. Also has a consistency that is good for a wide range of cooler types, so gets a lot of recommendations. Direct die to somewhat smooth copper, should work fine. I've used it on GPUs in the past.

 

[tilophase]

Takes a while to cool down those heat sinks after they get warm, they are designed for those short bursts that the majority of users experience, not continuous.

That's what I would have thought too. But right now, when idling, CPU is at ~60°C. The instant I hit it with a load, the temp jumps to high 90s and starts thermal throttling. There's not even a hint of that increasing curve that was present with the freshly applied thermal paste. Is there anything other than the thermal paste that could cause this behaviour?

 

[Eximo]

Aside from a damaged component, nothing really comes to mind.

 

[Habesan]

Are you sure you removed the dust from the radiator?

 

[tilophase]

Are you sure you removed the dust from the radiator?

Yes. I went over everything with a vacuum. The sudden spikes in temperature makes me feel like the overall thermal mass that is available to the CPU is almost non-existent, and I can't think of anything that could cause this other than bad thermal conductivity between the CPU and the heatsink.

Aside from a damaged component, nothing really comes to mind.

I'm assuming you mean a damaged CPU?

 

[Habesan]

What if the heat pipe has lost fluid ?

 

[tilophase]

What if the heat pipe has lost fluid ?

I didn't even know that was possible. Is there any way to verify that?

 

[Habesan]

The simplest test is to touch the warm water with one end of the heatpipe.
If the other end heats up very quickly, then the heatpipe works well.
You can also compare it to a solid copper wire of the same size or an empty copper pipe.

A slightly more complicated procedure is to place the heatsink on a heat source of known power.
For example, a 100W or 200W power rated resistor.
Actual power can be easily calculated from the applied voltage and resistance of the resistor.
see: Overclockers.com/ Heatsink Testing Methodology.