Question Strange experience with repasting my laptop

[AndersonJames_076]

Greetings everyone, I’d like to share my extremely strange experience repasting my HP Victus laptop and get some suggestions from you all.


I decided to use Arctic MX-6 as my thermal paste because of its high viscosity and better performance compared to MX-4. Initially, it worked well, reducing my CPU temps by 5°C and my GPU temps by 3°C, which made me happy. However, the next day, my performance started dropping. I had been running stress tests daily at the same time to check for pump-out issues, and unfortunately, that’s exactly what happened. My CPU, which initially averaged 48W at 99.5°C in a CPU-intensive test, dropped to 45W at 101°C the following day. A day later, I lost another 2W, and the CPU was clearly throttling.


Thinking the issue might be due to poor mounting pressure, I repasted again, but the same problem occurred within days, leading to performance loss. My GPU, on the other hand, remained stable at 85°C, just below its 87°C limit. Suspecting a bad batch of paste, I ordered another tube from a different seller, but the results were the same—throttling returned within two days.


Now, I’ve decided to switch to PTM 7950 since I’ve heard it handles pump-out better. However, I’m still surprised that even a thick paste like MX-6 failed in my case. To confirm the issue, I checked the paste after every repaste and consistently found signs of pump-out. I also experimented with different application methods, including the pea drop and spread methods, and even used excess paste while ensuring all screws were tightened properly. Despite all this, the problem persisted.


At this point, I’d appreciate any insights on what I might be doing wrong and whether PTM 7950 is the best option for my situation.


Thanks for reading!

 

[CelicaGT]

What CPU/GPU? Also, every laptop throttles, there is just is not enough cooling capacity. The published wattage and clockspeeds are for burst workloads, not sustained.

 

[AndersonJames_076]

What CPU/GPU? Also, every laptop throttles, there is just is not enough cooling capacity. The published wattage and clockspeeds are for burst workloads, not sustained.

I have a Ryzen 5 5600H and an RTX 3050 Laptop GPU. On the first day of testing, my CPU was able to maintain 48W while running CPU Burner. To confirm that throttling was occurring, I ran the test again 20 minutes ago, and now it's down to 38W, it’s gotten even worse. This should not happen, it never happneed with the factory applied thermal paste from HP

 

[CelicaGT]

I have a Ryzen 5 5600H and an RTX 3050 Laptop GPU. On the first day of testing, my CPU was able to maintain 48W while running CPU Burner. To confirm that throttling was occurring, I ran the test again 20 minutes ago, and now it's down to 38W, it’s gotten even worse. This should not happen, it never happneed with the factory applied thermal paste from HP

First of all, you aren't helping matters by running constant stress tests, you can actually do physical damage with some so be wary. Second, too much/thick thermal paste can be as bad as not enough. Third, ensure you are tightening the thermal solution screws in the proper order, HP should have information regarding such. Fourth, in your previous testing with the OEM paste did you record fan speeds etc, were you running the same power profile, on battery? On mains? On the same surface? (Flat, stable surface. Some machines throttle aggressively when moved.). Fifth, on the matter of PTM 7950. I have seen it highly recommended, as it takes away some of the variability of paste applications and has decent thermal conductivity. When I redo my G15 (with some mashup of paste/liquid metal Dell uses) I intend to try it. That's probably the only endorsement I can give.

 

[AndersonJames_076]

First of all, you aren't helping matters by running constant stress tests, you can actually do physical damage with some so be wary. Second, too much/thick thermal paste can be as bad as not enough. Third, ensure you are tightening the thermal solution screws in the proper order, HP should have information regarding such. Fourth, in your previous testing with the OEM paste did you record fan speeds etc, were you running the same power profile, on battery? On mains? On the same surface? (Flat, stable surface. Some machines throttle aggressively when moved.). Fifth, on the matter of PTM 7950. I have seen it highly recommended, as it takes away some of the variability of paste applications and has decent thermal conductivity. When I redo my G15 (with some mashup of paste/liquid metal Dell uses) I intend to try it. That's probably the only endorsement I can give.

I understand, and I appreciate your response. Yes, I'm aware that using too much thermal paste can be just as bad as applying too little, I only did that once to be sure. Regarding the screws, I made sure to tighten all seven in the correct order as labeled on the heatsink. I first tightened them lightly, then fully secured them in the proper sequence.

Whenever I conducted a stress test, my fans were set to max, and I was running the performance profile in Omen Gaming Hub. I frequently switch between performance and quiet modes, so I made sure it was set to performance, as the quiet profile limits both the CPU and GPU to 25W. The laptop was also plugged into mains and placed on the same table during all tests.

Now, coming back to my main issue.....pump-out. Despite taking all precautions, the thermal paste still seems to pump out every time. My idle temps remain consistent, currently at 55°C, but under load, the temperature spikes instantly to 101°C with just 55W of power draw. On the first day of testing, it hit 99°C at 60W. Both 55W and 60W only appeared briefly before the wattage settled at 48W. However, when comparing screenshots from day one to today, the wattage drop is noticeably quicker now.

Interestingly, my GPU isn’t experiencing any issues, which surprises me. This makes me wonder if MX-6 just isn't suited for direct-die cooling. A typical desktop CPU wouldn’t spike to 100°C and stay there while gaming.

 

[CelicaGT]

I understand, and I appreciate your response. Yes, I'm aware that using too much thermal paste can be just as bad as applying too little, I only did that once to be sure. Regarding the screws, I made sure to tighten all seven in the correct order as labeled on the heatsink. I first tightened them lightly, then fully secured them in the proper sequence.

Whenever I conducted a stress test, my fans were set to max, and I was running the performance profile in Omen Gaming Hub. I frequently switch between performance and quiet modes, so I made sure it was set to performance, as the quiet profile limits both the CPU and GPU to 25W. The laptop was also plugged into mains and placed on the same table during all tests.

Now, coming back to my main issue.....pump-out. Despite taking all precautions, the thermal paste still seems to pump out every time. My idle temps remain consistent, currently at 55°C, but under load, the temperature spikes instantly to 101°C with just 55W of power draw. On the first day of testing, it hit 99°C at 60W. Both 55W and 60W only appeared briefly before the wattage settled at 48W. However, when comparing screenshots from day one to today, the wattage drop is noticeably quicker now.

Interestingly, my GPU isn’t experiencing any issues, which surprises me. This makes me wonder if MX-6 just isn't suited for direct-die cooling. A typical desktop CPU wouldn’t spike to 100°C and stay there while gaming.

Pump out (a little) is normal, as is the quick drop in thermal conductivity. It should level off after a bit of heat cycling. As for the CPU hitting 100°C that is fairly normal. FWIW tjmax is 105°C but throttling should begin before that. For gaming, you should rarely have all 6 cores pegged, so it should not be an issue but in my experience mobile CPU's tend to lag pretty far behind their desktop counterparts performance in practice, so you won't see those high clockspeeds they like to advertise.

Also consider that 60W is outside the configurable TDP for the 5600H, which is 35 to 54W with a default of 45W so you're actually doing pretty well. There is probably some timed boost stuff going on there where it can exceed the base TDP for a predetermined time for short bursty workloads, then settle back to a programmed default. Unfortunately I'm not terribly well versed on AMD mobile processor behavior, but this is how the Intel stuff works.

Here is the data page for the 5600H:

https://www.amd.com/en/support/down...yzen/ryzen-5000-series/amd-ryzen-5-5600h.html

 

[AndersonJames_076]

Pump out (a little) is normal, as is the quick drop in thermal conductivity. It should level off after a bit of heat cycling. As for the CPU hitting 100°C that is fairly normal. FWIW tjmax is 105°C but throttling should begin before that. For gaming, you should rarely have all 6 cores pegged, so it should not be an issue but in my experience mobile CPU's tend to lag pretty far behind their desktop counterparts performance in practice, so you won't see those high clockspeeds they like to advertise.

Also consider that 60W is outside the configurable TDP for the 5600H, which is 35 to 54W with a default of 45W so you're actually doing pretty well. There is probably some timed boost stuff going on there where it can exceed the base TDP for a predetermined time for short bursty workloads, then settle back to a programmed default. Unfortunately I'm not terribly well versed on AMD mobile processor behavior, but this is how the Intel stuff works.

Here is the data page for the 5600H:

https://www.amd.com/en/support/down...yzen/ryzen-5000-series/amd-ryzen-5-5600h.html

Thank you, my only concern was that if the pump out continues, the die will be left with no thermal paste and will become dry. I hope that doesn't happen.

Edit: I came across a few threads on reddit where people were saying that only their cpu pumped out while their gpu was doing fine. I think it supports my theory of MX 6 not being a good paste for direct die cooling. I'm not sure how ptm will perform consider that MX 6 was a very viscous paste that pumped out. I can only hope and pray that ptm lasts

 

[CelicaGT]

Thank you, my only concern was that if the pump out continues, the die will be left with no thermal paste and will become dry. I hope that doesn't happen

It will not all pump out, and surprisingly little is required for good thermal conductivity. I install similar compounds on large industrial motor drives. There is some initial "pump out" of excessive compound after some heat cycles. Over time (years, typically) the carrier liquid will dry out. This is also normal and the reason for repasting. If the thermal solution gets disturbed physically it will break contact between the cold plate of the heat sink and thermal load (CPU/GPU). The lack of liquid stops the remaining dry compound from reforming a good contact. Extreme heat cycling (like in a laptop or desktop CPU application) of dry compound can also cause a break in contact due to the dissimilar expansion rates of the load and cold plate. For this reason many OEMs are now using a thermal putty or phase change type materials that do not dry out. Nobody likes warranty repairs.