Uneven CPU thermals!

So can someone explain if Framework did any heat/thermal/throttling testing before offering the machines from sale? Is there a design issue and if so why wasn’t it caught during testing for thermals before approval for manufacturing? Do all these individual test results being posted mean our Framework machines are defective?

Technically not defective due to the fact that they promised cooling up to 45w and it appears to do that most of the time, however the cooler really should be capable of more so we’re trying to overcome the flaws in the design to allow the CPU more headroom.

I have the same product from Thermal Grizzly and I can’t reach those scores. I actually just disassembled, cleaned, and reapplied to see if I messed something up somewhere… Have you done any other optimizations with software or undervolting?

Maybe Framework needs to partner with them on the AirJet cooling.

1 Like

While in theory yes you could use those to cool a laptop like this, it would require so many of them that there’s no way it could be cost effective. It should be quieter though which would be nice.
I’m pretty sure those are just designed as an fix to the passively cooled BS that companies like apple are putting out these days.

1 Like

Would a smaller form factor 2230 SSD be better for heat management. Although a high performance 2230 like Crucial P310 or Corsair MP600 MINI may still generate heat but being physically smaller and taking a lot less real-estate in the SSD slot may help with airflow in an already limited space?

I have both NVME populated, they only heat up on intense tasks, those are not that frequent, and I don’t see any impact. Choose either, both and whatever performance your budgets allows, overheating is no concern.

2 Likes

no absolutly not. There is no Airflow in the SSD Area, its completly cooled to the Thermalpad on the lower Case and the Midplate. The SSD Temps are Mint on this Device. Those little Watts have no Way to transfer into the HEatsink.

1 Like

I see, so you are of the opinion the 2230 form factor has a disadvantage for heat dispersion since they cant fully make contact with the thermalpad and midplate? And a full size 2280 SSD would make better contact with the thermalpad and midplate?

I would imagine taking up half the space as a 2280 would allow more room for heat to move/dissipate away from the SSD in the extra space around it in the socket that would be otherwise occupied be a full size SSD module? … maybe I am mistaken.

Huh? When did i say this? I am running double stacked. 2230 for OS and 2280 for Storage. Reread what i have written.
I answered your Question if a 2230 is better for Thermal Management. And answered it with absolutly not, there is no difference as the Heat of the lower SSD Slot is dumped into the Lower Case and the upper slot is dumped into the Midplate. There is no Contact to the Heatsink.

1 Like

There isn’t any airflow in that area. And air is a basically an insulator. It is terrible at conducting heat. There’s no airflow to block and there’s surely no advantage to leaving air space, rather than taking up that space with an SSD. And as PSierra117 said, there are theremal pads for both the 2230 and the 2280 SSD to transfer heat from them to the bottom case and the midplate, respectively. So there’s no thermal advantage to leaving extra space.

1 Like

That test was using the PTM7950 from Aliexpress.

Now that I had the “Thermal Grizzly PhaseSheet PTM” for 24 hours, I decided to run some tests.


CPU Overclocked - OFF

Single Run, 16581 pts

10 Minutes Test, 16708 pts


Tests with the CPU Overclock from the AMD Adrenalin.

CPU Overclocked - ON

Single Run, 16636 pts

10 Minutes Test, 16728 pts

I like the Thermal Grizzly PhaseSheet more as it feels a lot more stable compared to the PTM7950 from Aliexpress.

Out of curiosity, how many pts do you have for a single and the 10-minute run?

1 Like

@PSierra117 @Alex_Uta What would be the optimal course of action for someone like myself (mainboard already replaced but performance degraded again)?

  • Replace existing LM with PTM7950/TG PhaseSheet
  • Remove pre-installed shim and install one of my own (along with adding PTM7950)
  • Buy new Heatsink assembly and use pre-applied LM
  • Buy new Heatsink assembly and use PTM7950/TG PhaseSheet

I’m currently leaning towards trying the first one just to see what happens, but I wanted y’all’s opinions since y’all have the experience and data to give a recommendation.

So far I had good results with PTM7950 (fake) and TG PhaseSheet.

I’d say that it will be worth just replacing the current LM with one of them.
I don’t have access to the real PTM7950 to test, but the TG PhaseSheet looks promising so far.

Just ensure you’ll properly clean the heatsink before installing it back. If you have access to a hot air rework station, I’d even go as far as trying to heat the heatsink up to 55/60 to make sure that there’s no LM left.

2 Likes

Out of curiosity, how many pts do you have for a single and the 10-minute run?

I just tested and even with the laptop propped up in the back to try to increase airflow I’m only getting a single run score of 15997 and 10 min score of 15544. Ambient is around 68-69F.

Currently I have a single layer of TG PhaseSheet applied.

Are you also running yours with the GPU module in? I notice that a great deal of the airflow goes through the GPU cooler rather than the CPU cooler.

Edit: I had some free time tonight so I tried taping off the back of the GPU module just to see if it actually did make a difference and it appears to have made little to no difference at all.

I then put some work into undervolting the CPU and managed to get to 16400 on the 10 min test by focusing on core 4, which seems to be running quite a bit hotter than the others, even after a couple of re-applications of PTM.

I’m pretty much done messing with it for now though. Once I get the liquid metal in I will try that out as well.

That’s strange as I had good results with a single layer of TG PhaseSheet.

Yup. I’m using the GPU module.

That could work, but, there’s something wrong either way.

Sounds like a plan :slightly_smiling_face:

Also, there’s one extra small detail. I’m currently using a new heatsink, so maybe framework fixed part of the issue with the new heatsink.

A quick test from cold but, with bad airflow resulted in 70.825W and 16599 pts.

Test

Well, if a new heatsink will fix the issues, I hope that Framework will provide us a test-bed to test the current setup, and if it fails, provides us a replacement heat-sink.

2 Likes

That worked for me.
New heatsink and Thermal Grizzly PhaseSheet PTM :slightly_smiling_face:

I hope that soon we’ll have an official answer to this but, I have a feeling that it will take months and most of the laptops will be out of warranty.

3 Likes

Legal warranty in europe is 2 years. I’ll keep an eye on the process :slight_smile:

2 Likes

Yeah even burst package power can’t reach over 65w (normally not even 60w) and continuous draw over the 10 mins drops to around 49w. If the new liquid metal doesn’t improve things to a sufficient level I’ll have to order a new heatsink.

I just want to get the performance out of the chip that it should have while also having the cooling overhead that would allow me to step back the power limit a little and allow quieter operation when needed.

1 Like