Is this Beta BIOS available by now? I just upgraded mine and started to have this issue and would love to get it fixed somehow.
Not yet
That sucks because my FW16 became pretty much useless due to this. I disabled auto suspend, but this issue still happens after a period.
I’ve found the issue is far less likely if I just shutdown when I’m not using the device. If that’s something you’re able to do, perhaps that can help in the mean time.
1 Like
It happened today on my fw16. I fixed following @Zac_Bowling “EC Reset“ solution.
It happened during a intensive cpu task (with whisper) and with accidentally disconnecting the 180W power adapter.
There’s a beta BIOS update that was released today for the AMD AI 300 boards which purportedly handles one particular “600 mhz” issue. I don’t know that it addresses what you’re experiencing - there are notes in the forum post that should be instructive on that topic.
Post: https://community.frame.work/t/framework-laptop-16-ryzen-ai300-bios-3-05-release-beta/
Thanks! I have the Ryzen 9 7940HS.
Ah dang - noted. This thread is marked for the AI 300 boards so I was thinking you were rocking one of those.
I honestly have not seen the 600MHz issue on the 7940HS. There is a very similar (probably same cause) 544MHz issue we see a lot on the 7040 boards…..but again, have not yet seen this 600MHz one.
It is 544MHz because that’s the minimum frequency of the cores in this processor. But the root cause is likely the same, for some reason the chip is throttled to the minimum.
I will move my convo to the appropriate topic though. I didn’t check the tags and this one came up when I searched.
i guess i find core of a problem, or at least have reliable way to reproduce this issue on my hardware. Below is what Claude summarized on this issue:
Root cause identified: iGPU VRAM bandwidth pressure triggers EC hard power-off /
PROCHOT lockSystem: FW16, Ryzen 9 7940HS, Radeon 780M (iGPU) + RX 7700S (dGPU), BIOS 4.03,
56GB RAM, Linux 6.18.19 (NixOS)I’ve been hitting the 544 MHz PROCHOT lock intermittently for weeks
(post-suspend, during gaming, etc). EC resets and BIOS 4.03 helped but didn’t
eliminate it. I was able to isolate and reliably reproduce the root cause.The trigger: active iGPU memory bandwidth saturation.
I wrote a minimal Vulkan reproducer that allocates 2GB DEVICE_LOCAL on the 780M
and does continuous vkCmdCopyBuffer ping-pong between two 1GB buffers:
vram-stress.cResult: instant hard power-off within seconds - system cuts power as if AC was
unplugged. No kernel panic, no OOM, no thermal shutdown in logs. Reproduced
twice consecutively. Note that BIOS 3.05 beta does not fix this either.This is the severe form of the same bug. The milder form (544/600 MHz PROCHOT
lock) happens with lighter iGPU memory pressure - in my case, a Wayland
compositor shell (quickshell) rendering ~27 surfaces on the iGPU.Journal evidence:
- Crash boot last log entry: normal userspace activity at 13:18:20
- Next boot: cold start at 13:19:00
- Zero kernel messages between - EC killed power with no software-level warning
- Earlier passive allocation test (2GB allocated but idle) did NOT trigger it -
the memory must be actively usedRelated firmware bugs on every boot:
ACPI: thermal: [Firmware Bug]: Invalid critical threshold (-274000)
ACPI: thermal: [Firmware Bug]: No valid trip points!
Thermal zones 4-7 have no valid trip points configured in ACPI tables. This may
explain why the EC’s response is so extreme - missing thermal protection
thresholds could cause the EC to use a blunt power-cut fallback instead of
graceful throttling.Workaround that eliminates the issue completely: Force GPU-heavy processes to
the dGPU with DRI_PRIME=1. Moving my compositor to the RX 7700S made the 544 MHz
locks disappear entirely.Why this matters for the thread: Everyone here is seeing the same PROCHOT
assertion from the EC, but triggered by different iGPU load patterns
(suspend/resume GPU state restoration, driver installation, gaming). The common
factor is the 780M’s shared memory controller hitting a power delivery limit
that the EC can’t handle gracefully.
I have tried the “vkstress.c” but it does not can a reboot here.
How exactly do you cause a forced reboot using it?
when i try to launch it on my machine it just goes blank, as if there is no battery and you pull out AC charger. I also had 544 mhz issue with games when cpu get heated or is memory in iGPU gets closer to 2gb used
We (users) have reports on this community forum, of forced reboot. But not forced power off.
I think you have a hardware fault, so best to raise a ticket with FW support via their web form.
So, I think your problem is unique and most likely hardware related.
I am just another user like you.
Have the same problem. New laptop.
So far happened once. The symptopm is exactly the same: wake up after a long suspend and cpu is in a “600 mhz pstate”. Not 544 guys, exactly 600.
Re-plugging AC helps. Nothing else helps. Tried to re-kick cpufreq, boost, etc. Which suggests it’s a hardware/bios problem. Doesn’t happen on Lenovo P14 or Asus ProArt 16 with the same or similar CPU.
System firmware is at 0.0.3.5