I think I have a VERY similar issue, exactly as if someone pulled the power cord. Started happening back at the end of May for me. Haven’t been able to use my 128G Max+395 FW Desktop since, and struggled with support.
Please read my other reply for details.
my first suggestion is to install some kind of app to sense the TCTL temperature. my understanding is that tctl should not exceed 100c. in my case it did once for very short time like 2seconds. which lead me to investigate the thermal paste inside.
Unfortunately I can’t even get it to stay on long enough to boot, let alone install software. I’ve tried live image, network boot, USB boot, ubuntu on ssd.
I was monitoring the temperatures before all of this happened and the max of all temps in the system was 86 degrees (I think that was GPU temp during gaming), and as soon as I closed games it would return to ~35-40 degrees within 5-10 seconds I seem to recall (it was pretty fast to cool back down). Unless the application of thermal paste from Framework was faulty or the thermal paste itself is faulty, I don’t really see why this would be happening.
1 Like
Hi, after a couple of days some News from my side:
- Tested a 1000W ASUS TUF Gaming PSU (documented to tolerate >140% rated current / >125% rated power for several seconds before cutting off): no change, shutdown still occurs. This makes a PSU/overcurrent-protection explanation increasingly unlikely.
- MemTest86: Passed
So far - so boring.
Exciting new finding: cooling strongly affects crash frequency.
- With a fan/AC blowing cold air directly into the open case, I can run the Far Cry 5 benchmark 10 times before a shutdown occurs (vs. failing on the very first attempt with the case closed).
- Temps recorded during FC5 (GPU power draw identical in both cases, 140W):
- With cold air: CPU ~67°C, GPU ~63°C, SSD ~28°C
- Without (closed case): CPU ~76°C, GPU ~75°C, SSD ~40°C
- -> But are that +10 Degrees really the decisive factor?!
Timing also matters:
- Starting the FC5 benchmark immediately after launching FC5 the system crashes within 0–5 seconds (0 seconds = immediatelly after pressing F5 so start the benchmark),
- whereas waiting 10–20s in the game’s main menu first delays the crash significantly, sometimes the run completes fully.
- Obersvation: launching the game itself causes a brief CPU spike to 95°C before settling to 70–80°C — so starting the benchmark right away stacks it on top of that initial heat spike. No idea whether that is important or not…
So do I have a thermal problem?
- Counter-test with Prime95 (CPU-only load, no A/C): CPU reached 100°C, held there for a few minutes, then throttled down gracefully to ~80°C (power draw also dropped, ~120W). No shutdown over 20 minutes.
- So it doesn’t look like a simple temperature threshold (100°C under Prime95 was handled fine via normal throttling) — it looks more like elevated temperature increasing the odds of hitting whatever GPU power-state-transition issue is at play, rather than heat being a trigger on its own.
Need to keep in mind: Those Shutdowns occur even during normal Windows operation without any load and without graphics-intensive applications, though they are rather rare (about once a day, sometimes even less frequently). In my opinion, this does not point to a thermal issue, since the system is so cool that you can’t even hear the fans.
But - I have heard rumors (and read it somewhere in this forum) that sometimes the thermal paste is not applied correctly when the board is shipped, and/or that the protection foil between heatsink and CPU has not been removed. Can I somewhat see it “from outside” whether my board is also affected without having to remove the heatsink?
That could be a thermal problem. The CPU shouldn’t hold at 100C for more than 30 seconds, let alone a few minutes.
i removed all my power bootup settings, then using “ryzenadj -i”, this is the “stock” boot up setting. tctl is default 100C. which is why i think on day 1, i could hit 102C.
since we already read about large burst draw of current/power by strix apu, my logic would be to stem any sudden overshoot = i set TCTL to 98C or even better 85C. at 98C, it still can overshoot to 100C, so there is an actual lapse in detection, thermal mass lag etc.
amd flaw? thermal paste flaw? i do not experience the blackouts, but i ended up setting a low tctl setting, and yes i did change the paste.
CPU Family: Strix Halo
SMU BIOS Interface Version: 25
Version: v0.19.0
PM Table Version: 64020c
| Name | Value | Parameter |
|---------------------|-----------|--------------------|
| STAPM LIMIT | 100.000 | stapm-limit |
| STAPM VALUE | 8.612 | |
| PPT LIMIT FAST | 115.000 | fast-limit |
| PPT VALUE FAST | 15.796 | |
| PPT LIMIT SLOW | 100.000 | slow-limit |
| PPT VALUE SLOW | 8.848 | |
| StapmTimeConst | nan | stapm-time |
| SlowPPTTimeConst | nan | slow-time |
| PPT LIMIT APU | 70.000 | apu-slow-limit |
| PPT VALUE APU | 0.000 | |
| TDC LIMIT VDD | nan | vrm-current |
| TDC VALUE VDD | nan | |
| TDC LIMIT SOC | nan | vrmsoc-current |
| TDC VALUE SOC | nan | |
| EDC LIMIT VDD | nan | vrmmax-current |
| EDC VALUE VDD | nan | |
| EDC LIMIT SOC | nan | vrmsocmax-current |
| EDC VALUE SOC | nan | |
| THM LIMIT CORE | 100.000 | tctl-temp |
| THM VALUE CORE | 35.670 | |
| STT LIMIT APU | 100.000 | apu-skin-temp |
| STT VALUE APU | 35.670 | |
| STT LIMIT dGPU | 100.000 | dgpu-skin-temp |
| STT VALUE dGPU | 36.307 | |
| CCLK Boost SETPOINT | nan | power-saving / |
| CCLK BUSY VALUE | nan | max-performance |
How did FW support respond?
As the psu is not the problem, and it happens when the cpu is not doing much, i would ask for a replacement mainboard. It takes just one bad solder joint to cause problems like this.
Also, not many other people reporting similar problems.
When running Prime95, the value for “CPU-Core” in HWiNFO jumps to 100C, remains for few (5–6) minutes, and then slowly decreases to approx. 80C after ca. 10 minutes.
I assume that this is not normal? What is the behavior of your system?
Yes I am in contact with the Support Team. I provided basically the same information as in this community topic, and they came to the conclusion that the mainboard should be replaced.
I am quite happy with the fast response times of the Support Team and now I hope that the new mainboard (which should arrive in a few days) will work stable…
Today I got the replacement Mainboard. And guess what… drumrolls… the problem is gone. No more shut-downs. Tried FC5 Benchmark-Test a couple of times as well as some more ones like Timespy while Windows is set to “Best Performance” - all good, no crashes.
BUT there is one notable difference: The replacement Board was shipped with BIOS Version 3.04, and I did NOT update to 3.06 (yet).
Well I assume that it is a hardware defect, but… what are the chances that it ultimately comes down to the BIOS instead of a hardware problem…?
What’s the temperature? Does it run cooler than the problematic mainboard?
Fancy rolling the dice? 
Yes and no:
- FC5 Benchmark: Same tempreatures as before, around 75 Celsius. That’s somewhat normal, isn’t it? BUT one notable observation: That short peak to 95 Celsius right when starting the game is gone. Tried several times. Maximum according to HWiNFO is 80 Celsius!
- Prime95: Still shoots up to 100 Celsius and stays 100 for few minutes and then slowly gets “cooler”…
So maybe the cause on the defective Board is a thermal problem with a component that is not monitored by a sensor and shown in HWiNFO…?
Nope, never change a running system 
But I also asked the Support Team in how far the BIOS Version might be the real problem…
1 Like
wow prime95 still going 100C
I just ram prime95 (package name mprime) on my Framework Laptop (not the Desktop OP has) and my CPU went straight to PROCHOT and clock speed went down to ~1GHz. I think this thing stresses a computer more than other stress tester