Using stress-ng doing ram stress test over 80% ram. I cannot get the Ram over 50C yet. I will run the test for longer to see if i can get it hotter.
Is there a setting in BIOS (including Smokeless UMAF) that makes the computer ignore it?
nope. Bios is very limited. Unfortunely i think i will return the machine. Too bad, i loved it but if i canāt fully use it itās uselessā¦ I even bought the RTX 5070 module. Hope the RMA process is not too bad
So it looks like you are just experiencing the same errant PROCHOT trigger the rest of us have also had since this laptop was released a year and a half ago.
Itās why my games and mangohud always think it is throttlingā¦. Allā¦. Theā¦.. Damnā¦. Time.
That said, I donāt think it is really throttling in many of these cases. At least performance seemed to be fine on past bios versions. I canāt say the same for 4.03 though.
Not really lol. I just closed my eyes to it (stopped looking at mangohud, for example), and just enjoyed my games.
1 Like
For full transparency: Yesterday, after cleaning up everything, I requested a return, and the support team has since escalated to RMA. I will continue to update this thread for the sake of completeness.
So sad but this is life..
1 Like
The support team already provided RMA labels (1 working day as expected) despite initially warning of potential delays in the process. Thatās a good news!
I hope that in a few years, when Framework Laptops have matured sufficiently, I may purchase another one from them.
Yeah I had a similar issue, I swapped out my 1st gen 7700s GPU for the 2nd gen and the PROCHOT went away, so I expect it was really throttling because of the GPU.
I wish there were a way to actually check what the cause is. Hopefully Framework can add this to the EC console output.
Just to provide an update, the package arrived at the facility in Germany on January 22nd, and today I received a full refund without any restocking fees or additional charges.
The refund was issued on January 24th and the entire experience was seamless.
I am truly delighted by the exceptional customer support and hope to become again a customer of theirs once again in the future.
I have the same issue, at least regarding performance wise. I am playing a game at 45FPS and suddenly the fps drops down to 20, while a few times in a second producing 40 only for a brief moment, as I found out the culprit can be the stapm limit, as my ram sticks were hot( 70, 78 C), I donāt think that was the issue. When I remove the charger the prochot signs inside amdgpu_top goes away, however the performance doesnāt increases. I am running the igpu.
+1 I can replicate and confirm, the moment one my RAM sticks hits 80C I start throttling. There appears to be a ~5C difference between the two, so theyāre not being evenly cooled if nothing else.
Even if the fans are set to 100% duty cycle/full speed, the CPU and GPU will not hit their thermal limits while the RAM will. CPU only stress tests will not trigger this, it needs to be a workload that makes heavy use of the RAM as VRAM I think.
This does appear to be a hardware problem that requires a hardware solution. Either additional chassis only airflow or some way to tie the RAM into the existing heatpipes.
1 Like
Is 80c a threshold for RAM throttling? I believe I may have that same phenomenon as you, but it didnāt stand out to me as concerning when I was watching temps. I think one of my sticks was at about 55c and the other 75c. As far as Iām aware, thereās no way for RAM to throttle, as itās timings are fixed (but I could be wrong). Granted, Iāve seen my infinity fabric clock speed change, but Iām not sure what that means. Iām using: G.SKILL Ripjaws SO-DIMM 32GB (2 x 16GB) 262-Pin DDR5 SO-DIMM DDR5 5600 (PC5 44800) Laptop Memory Model F5-5600S4040A16GX2-RS - Newegg.com
If thatās whatās happening to me though, it would certainly explain some things.
Update: I wasnāt able to find any online references saying 80c is a throttle threshold for ram. I did find a reddit post talking about crash at temperatures >110c though.
Of the tech specs Iāve found for DDR5 SODIMM modules, it looks like thereās generally 85c-95c listed as the max safe operating temperature. I canāt find that number for your type of ram right now, though - it seems like not all manufacturers/retailers keep a public datasheet where I can find it (ā¦or Iām not looking well).
Reportedly [forum posts and articles Iām pulling up], RAM can throttle due to high temps. Allegedly most DDR5 SODIMM modules will have some on-die temperature sensor, generally built into the SPD hub on the module, which can report out to the board carrying the RAM.
Micron doc talking about it on page 2: https://assets.micron.com/adobe/assets/urn:aaid:aem:5902f17e-8bde-4246-971c-8d7f274c6b1a/renditions/original/as/ddr5-key-module-features-wp-client.pdf
Some user posts online seem to suggest RAM modules themselves may be able to perform their own down regulation when experiencing high temps, but I havenāt found a trustworthy source for that today.
2 Likes
Thanks for the info! I looked myself as well and couldnāt find a spec sheet for that sku either. I do know that it has a temperature sensor that reports out because Iāve seen the output. Iāll do some testing tonight to see if when I encounter my overheating issue if itās correlated with RAM temperature.
1 Like
Looking at a thermal image of mine from another thread, it does look like a part of my ram stick (top right corner) could be getting overly hotā¦ unfortunately the actual temp-reading wasnāt on that location directly.
Although I no longer possess the machine, I can confirm that a CPU stress test does not provoke the issue; it must be a workload that stresses both the CPU and the RAM modules.
In my case i used OCCT with the CPU+RAM stress test configuration
Conor got to it before I did, I also could not find anything official specifically delineating 80C as a notable temperature. There are forum posts mentioning it here and there (like this Level1Tech post) but the general consensus seems to be it should be running cooler (~70C) to avoid problems.
I did find dead links to Crucial documentation, where the poster mentioned 95C as a significant (bad) value called out by Crucial, but I canāt find a live page saying anything from Crucial on the subject matter. Itās probably on the waybackmachine, but I donāt see any reason to doubt it since other manufacturers seem to be around the same ball park.
There is some question about whether or not the sensor values themselves are right, Iāve seen some posts mentioning a 10-15C delta between actual temp and sensor values, and that would put it right into the danger zone. Except I already have two different āRAMā sensors with that kind of value gap: ACPI sensor #3 (called F75303_DDR in the fan curve tool Kemal_Ozturk released) reports about ~10C difference from the SPD sensors on the I2C bus.
There were reports online of the VRM in particular being the pain point for DDR5, I can easily see a situation where the actual RAM sensors arenāt yet too hot, but the VRM is well past it and the voltage instability starts causing timing issues. It also seems like VRMs in general donāt tend to have a dedicated sensor that can be read, so they either drop a warning in the i2C bus or you just just get degraded performance/systemic instability/dead components if theyāre regularly overheating.
Thanks for the info! I was able to do some testing and found that my temp sensors report a max of 72c on ram. Even if there are parts of the ram that are operating higher, the controls logic can only go based on whatever the temperature sensor says, so I donāt think this is the root of my issue. Learned a lot today though, so thanks!
Wait, I think I glossed over a very important detail on my first read through. Are you saying that what HWInfo64 reports as TZ.TZ03 is a ram thermal sensor? That sensor Iāve seen get very hot, near 100c.
TZ03 is the CPU temperature, same as Tctl. For RAM thereās a sensor named spd5118, one for each stick
1 Like