FW16 (AMD 7040 / Ryzen 7840HS) + RX 7700S: dGPU seems capped to ~35–40W under sustained load on Linux (firmware 0.0.4.2) – similar 35W cap reports on Windows too

Community Support
,
1

Which OS?

  • Linux

Which release of your OS?

  • Arch Linux (rolling, fully up-to-date at time of posting)
  • Desktop session: KDE Plasma Wayland

Which Framework product and generation?

  • Framework Laptop 16
  • CPU platform: AMD 7040 Series (Ryzen 7 7840HS)
  • dGPU module: AMD Radeon RX 7700S
  • RAM: 64 GB
  • Firmware: 0.0.4.2 (issue still reproducible)

Summary / impact
On my FW16 under Linux, the RX 7700S appears to be hard-limited to roughly ~35–40W under sustained GPU load. Even when the dGPU is clearly active (e.g. a Steam/Proton game launched with DRI_PRIME=1; radeontop shows the dGPU near 100% utilisation and the iGPU relatively low), real-world performance is often nearly identical to using the 780M iGPU.

My primary reason for buying the RX 7700S module is local compute on Linux (LLM inference / LoRA training etc.). If the dGPU is silently capped at ~35–40W, users can waste significant time because workloads run far slower without an obvious “error”.

Charger / power context

  • I am currently using a third-party USB-C 140W adapter (not the Framework 180W adapter).
  • However, this does not look like a simple “adapter wattage too low” explanation:
    • There is an existing community report of a very similar ~35W cap on Windows.
    • There is an existing GitHub issue about not reaching full dGPU power even with a 240W PD 3.1 charger.
      So this may be a broader EC/firmware/PD negotiation or power-policy issue rather than just a specific charger.

Data from my system

  1. Power under load (uW):
    $ cat /sys/class/drm/card*/device/hwmon/hwmon*/power1_average
    29000000
    39255000
    = ~29.0W and ~39.3W. In my testing, the dGPU does not exceed this range.

  2. Clock tables (two devices: iGPU + dGPU):
    $ cat /sys/class/drm/card*/device/pp_dpm_sclk
    0: 255Mhz
    1: 794Mhz *
    2: 2208Mhz
    0: 800Mhz *
    1: 1100Mhz
    2: 2700Mhz

Related reports (please read)

Questions (requesting data points)

  1. Linux users: can your RX 7700S exceed ~40W under sustained load on AC? If yes, what exact conditions make it work?

    • distro + kernel/Mesa
    • charger (Framework 180W vs 240W PD 3.1 vs other)
    • power mode / charging state / battery % (if relevant)

  2. Windows users: have you seen a similar ~35W cap? If yes, under what charger and power mode?

  3. Does Framework have an official statement or known-issue status for the ~35–40W cap behavior on FW16 + RX 7700S?

I’m also contacting Framework support with the same details. If Framework requests specific tests (different charger, supported distro, particular logs), I will update this thread with the results.

Christoph

2

please update to 4.03 beta, it should fix this issue. Framework Laptop 16 Ryzen 7040 BIOS 4.03 Release BETA

3

Please update to the beta bios release (4.03), confirm your issue, and then submit another GitHub issue for it. Getting Framework to acknowledge and fix these issues have been ongoing for over a year.

They keep saying they “fixed” it but they never do. All us 16 owners are suffering.

4

I hope that this new beta mentioned by others fixes it. I ended up returning my laptop over this issue. Framework has not ever officially acknowledged this issue publicly…. I just asked them about this on twitter a few weeks ago and they simply kept dodging the question / changing the subject until they just didn’t reply at all.
The short story is that everything will be capped to basically 65W total system power (100W including charging the battery) unless you use the official 240W Framework charter. No other charger will work properly even if that charger is capable of delivering more watts then you currently need.
The firmware seems to be incapable of managing power usage based on power adapter PD delivery values. It is basically a <240W = 65W power cap code issue. No logic is done whatsoever for any lower power chargers (atleast it wasn’t when I got the laptop, and also wasn’t the last time I looked at the firmware in 2025).
I really do hope they finally do something about this, because it makes an otherwise great laptop basically totally gimped.
the other thing they needed to fix was having some sort of fan controls whatsoever. I couldn’t leave this laptop on in my bedroom because it would wake us up at night randomly running the fans to 80%+ meanwhile it would let the CPU get very toasty sometimes before spinning the fans up.
I hope that is atleast fixed :slight_smile:

Good luck!

1 Like