Yes, I bought the DIY edition of the Framework 13. The bezel issue has been resolved now. After spending about an hour trying to make it fit, I eventually gave up. However, it seems that over time (maybe due to the pressure from being close), it finally stuck in place. So, the bezel problem isn’t as terrible now, but it was frustrating as a first impression with the DIY edition.
Regarding the CPU temperature, I haven’t done anything intensive with the laptop. I’m a developer by profession, but I haven’t been using the laptop for any professional work yet. So far, I’ve mostly just used it to browse the web. The fan noise and temperature spikes happened while I was watching some videos online. For monitoring, I’ve been using htop and sensors to check CPU activity and temperature.
I’m not really a hardware expert, but I wouldn’t expect such a powerful CPU to get so hot just from watching videos. That said, this issue seems to happen randomly. For example, right now, as I type this, the laptop is cool and the fans are quiet.
Which browser are you using and have you setup it to use VA-API?
Atleast Firefox by default has it disabled (this can depend on your OS - not sure if Ubuntu enables it by default for AMD). I don’t know if Chrome based browsers currently have working VA-API support - haven’t really followed up on it, but it has been real hit and miss previously (might work on some version and broken after a update etc)
I have the 7840U version and running CachyOS (Arch) and my fans are mostly quiet. Although I haven’t really stressed the system, only when building few AUR packages I can hear the fans starting to ramp up.
I am using Firefox, VA-API is enabled.
Also I have noticed this only while the laptop is attached to my monitor which I also use as a power source.
Considering those CPU issues appear randomly i have little information about it, but will keep the post updated once i gather more info and provide some outputs from the tools i am using.
power supply: yes
updated: So 15 Sep 2024 16:36:31 CEST (27 seconds ago)
has history: no
has statistics: no
battery
present: yes
state: charging
warning-level: none
energy: 48,6072 Wh
energy-full: 60,4804 Wh
energy-rate: 20,3872 W
charge-cycles: N/A
time to full: 34,9 minutes
percentage: 80%
icon-name: ‘battery-full-charging-symbolic’
monitor is DELL P3221D if that’s relevant, connection is made through usbc
In my experience laptop fans tend to be noisier when charging. All that additional heat inside the chassis.
I have a FW13 AMD and the fans are usually pretty quiet but do pulse on sometimes. I would like a quicker ramp to a quiet speed and a slower ramp to a loud one. I have not yet made any adjustments to the fan curve.
There is a lot of power on tap from the CPU but if I am compiling something with all cores max load the fans might start going. Are you running on balanced power plan still?
In Firefox does hardware decoding show a user config thst is forcing enabled? You can also use amdgpu_top to confirm gpu usage when playing video.
With max load any laptop with this CPU is probably going to be loud. What kind of cpu usage have you been seeing 100c at? Could be poor thermal paste or not ideal cooler alignment if it’s jumping up to 100c really easy.
When using another AMD laptop, the Lenovo Flex 5 14", I notice that when plugged in, due to it switching to the high performance turbo mode, the laptop gets quite hot pretty fast.
The turbo boost mode lets the laptop turbo quite aggressively until hitting either thermal or power limits.
If you are having issues, then try turning turbo boost off. My average temps without turbo boost enabled (on battery) are around 50C while once turbo mode is up it jumps up to the high 80s.
When the laptop is plugged in, the OS automatically changes to the high power profile (with turbo boost enabled) which is likely why the laptop may get hot quite quickly.
Another possibility is a faulty cooling system, so if you want to, you can file a support ticket to check with Framework whether your cooler is faulty. Framework Support is quite busy so it might take some time before you get a reply.
Just keep in mind it’s very normal for laptops with these cpu to hit 100c and then throttle back. They are using as much thermal headroom as possible to allow as much performance as possible.
Maybe try to see if your perf numbers match notebookcheck benchmarks for your model.
I received my 13 last week. After a bit of distro hopping, I did a clean install of Ubuntu 24.04.1 yesterday. While I’d experienced a few surprisingly loud bursts of fan noise over the past week running Fedora 40 and Arch, tonight I installed Steam and decided to take Counterstrike for a spin.
With my FW 13 lid closed, connected to an external display at 3440 x 1440 and 75 Hz and no other apps running, after the shaders compiled and the game launched, my FW sounded like a jet engine. After 20 minutes of play, the bottom of the device was literally too hot to touch.
Granted, I’m coming from a MacBook Air in all it’s glorious silence and heat management, but surely this cannot be normal?
If you create X watts of heat you have to dissipate X watts of heat. If you can’t dissipate X watts of heat you have to create less heat. The Macbook Air is going to burst high wattage and then throttle down to under 35w total system load over extended workloads. The FW13 may be able to dissipate more heat with more noise as the different variants top out between 50-65w total system power at full load.
The bottom of the case is likely to get hot on either device with 100% load in my opinion. It may take longer if one device heat is isolated from the bottom of the case versus. Eventually one will see heat soak across the entire chassis.
Now I am now sure how big of a difference the 7640U and the M3 are in performance per watt at max load. If you look at the M3 air vs larger Macbook Pros you will see that the Air throttle down hard and the Pros sustain the performance over extended periods. The Air does not include a cooling fan and the processor can reach temperatures as high as 114c.
One can look at previous devices like the ill fated Macbook 12". Large swaths of them have stopped functioning, do not detect soldered on SSD, have weird issues due to the BGA on the passively cooled CPU/SoC. Some people hope to save theirs from the same fate by putting a thermal pad on the top of the passive heatsink connecting it to the bottom chassis.
Obviously how well the cooling system can take heat from and dissipate it out into air is going to modulate how loud the system is. It could be better or worse than other peoples FW13 depending on how good of a fit the cooler is to the die.
The other issue may be that by closing the lid you lose additional air normally sucked through the keyboard (not 100% sure but usual in such designs). When I look at the hinge/bezel area where the fan exhausts it seems to me that the closed lid has a way smaller path than when the screen is open. It’s like a narrow slit on on the bottom with a 90 degree path for hot air to exhaust. When the display is open it’s able to escape at a shallow angle under the screen towards the surface it’s placed on and in front of the screen. I suspect that given the same load, over an extended period, it’s going to be quieter with the lid open than the lid closed.
That being said if one would like to reduce performance to benefit from lower fan noise one could do so. Most of the time when playing games on laptops I make sure I put framerate cap on matching the refresh rate.
even if it is. normal i don’t think it should be for a machine that I’ve paid almost 2k eur.
Anyway I decided to return my laptop and that also is a very painful process if you are in Europe.
However i still want the company to succeed and grow so that we can have all what they promise but they are just not there yet i think.
Maybe if the prices were lower I don’t know, but that’s another topic.
Not sure if i should close this or not, since my original issue is resolved (by returning the laptop) but other might still experience the issues so i would prefet this to stay open.
Maybe with the fan heating is beacuse of Ubuntu 24.10.1. I was using Ubuntu 24.10 and it was pretty quiet, looks like FW might need to do some updating maybe?
Notebookcheck’s review is not neutral, Framework sent a tuned laptop to the reviewer, as shown here
Transparency
The selection of devices to be reviewed is made by our editorial team. The test sample was provided to the author as a loan by the manufacturer or retailer for the purpose of this review. The lender had no influence on this review, nor did the manufacturer receive a copy of this review before publication. There was no obligation to publish this review. We never accept compensation or payment in return for our reviews. As an independent media company, Notebookcheck is not subjected to the authority of manufacturers, retailers or publishers.
The reviewed laptop has 28W STAPM limit removed and the PPT SLOW limit edited to 35W. This leads to higher benchmark results compared to our laptops.
Note the low APU STAPM limit %(it should be 100% upon sustained load) and the CPU Package power(should be 33W if limited by PPT SLOW, or 28W if limited by STAPM
I repasted my FL13 AMD using PTM7950, and the CPU temperature never reaches 90C on pulse or sustained load
I don’t agree with your insinuation that the review is biased. It could be inaccurate if the manufacturer sent a golden or altered sample. I also doubt Framework sent a tuned laptop to the reviewer. Your citation only says that they were provided with a test sample by the company or a retailer of the company.
With Windows 11 power mode set to “best performance” my 7640U will run at 35W package power without issue for some time until APU STAPM kicks in and it goes to about 33W. With Windows 11 power mode set to “balanced” PPT SLOW limits it to 28W.
As you can see Notebookcheck sets their Windows 11 power mode to “best performance”. Did you do this in your testing? What do you base your claim on?
Mine thermal throttles until it ramps the fan up. After it hits 100% PPT SLOW at either 35W (best performance profile) or 28W (balanced) until APU STAPM hits later and reduces to 33W. Are you sure this is not what yours does to?
In the future I will be repasting with PTM7950 if needed.
I changed my TDP to 35W but my CPU reaches 100C, making the actual power 33~34W, that’s on factory cooling before me repasting PTM, but the reviewer only got 85C
You experienced STAPM throttling even on Windows performance mode.
Please open HWINFO and pay attention to CPU PPT FAST Limit, CPU PPT SLOW Limit and APU STAPM Limit. These values are in percentage, indicating the actual power divide by these power limit values.
When stressing CPU, you’ll get the first 100% on CPU PPT FAST Limit for 5 seconds, then a 100% on CPU PPT SLOW Limit.
As the chassis heats up, the APU STAPM Limit % goes up until it reaches 100%, at this moment your CPU package power will go down, the CPU PPT SLOW Limit % also goes down as your CPU is no longer capped by slow limit, it’s capped by STAPM instead.
In the screenshot on the review article,
notice that the APU STAPM Limit % is very low, as low as CPU PPT FAST Limit %, this indicates that the STAPM is completely turned off during the review, which contradicts both your and mine observation.
Over what length of time? Were you heat soaked already before the test? What’s your ambient temp? If it heat soaks it’s going to go higher. The Prime95 stress test screenshot has a graph that shows the original boosted package power. Grid is 30 wide and I believe it is 10 seconds per square. That means 300 seconds. You can also see that their temperatures went up a lot higher into the 90s before settling down as the fan span up
I don’t agree. Look at their Prime95 + Furmark screenshot. It’s the very next one and it has APU STAPM at 100% and their package power is at 31W. It also looks very similar to the numbers I get from a cold boot launching Prime95 and watching hwinfo.
I’m surprised, how did you managed to do that? When my CPU skin temperature reaches 40C the APU STAPM starts to overtake CPU PPT SLOW Limit, reducing the package power as the result
The only change on the system is setting Windows 11 to Best Performance in the settings app. Otherwise it’s default running a prime95 torture test. It’s about 19c in the room and the laptop is sitting on a desk.
