Random hard reboots during heavy compilation on Slackware 15.0 - PSU aging or kernel panic?

[jalah93190]

Hi everyone,
I’ve been running Slackware 15.0 on my main daily driver for quite a while now, and generally, it has been the rock-solid experience I’ve come to expect from this distro. However, over the last two weeks, I’ve started running into a really frustrating issue that has me scratching my head, and I’m trying to determine if this is a software configuration issue or if my hardware is finally giving up the ghost.
The behavior is abrupt: the system performs a hard reboot without warning. No kernel panic on the screen, no freezing mouse cursor beforehand—just a sudden click and a restart.
I first noticed it while compiling a custom kernel. I usually run make -j12 because I’m on a Ryzen 5 3600, and about 10 minutes into the process, the system just cut out. My first instinct was overheating. I installed lm_sensors and monitored the temps while putting the system under load again. The CPU never went above 70°C, and the GPU (an older RX 580) was idling. So, thermals seem fine.
I dug through /var/log/syslog and /var/log/messages hoping to find a segfault or some error right before the crash, but there’s nothing. The logs just stop and restart. This lack of software logging is leading me to believe it’s a power delivery failure.

Here is where I could use some advice from the hardware gurus here. My rig isn't exactly power-hungry, but I have recently added three mechanical hard drives to set up a local RAID array for backups. The machine is currently powered by a generic 500-watt power supply https://serverorbit.com/power-supplies/ ... r/500-watt that I salvaged from an old office workstation years ago. It’s definitely not a "Gold" rated unit; it's just a standard workhorse I had lying around.
I know that capacitor aging is a real thing, and reliable power distribution is critical, especially when all cores are at 100% and mechanical drives are spinning. I suspect that the sudden spike in power draw during compilation, combined with the spin-up of the drives, might be tripping the over-current protection on this old 500-watt unit. It feels like the 12V rail just sags too much and the motherboard cuts power to protect itself.
Before I go out and tear my cable management apart to swap the unit, I wanted to ask:
Has anyone experienced similar "silent" reboots on Slackware that turned out to be software/driver related (maybe ACPI or C-states)?
For a Ryzen 3600 + RX 580 + 4 HDDs, do you think a modern, high-quality 500-watt unit is still sufficient headroom, or should I be looking at 650W+ to be safe?
I’d prefer to stick with a 500-watt unit if possible just to keep the efficiency curve tight for idle usage (since this machine runs 24/7), but stability is obviously the priority.
Thanks in advance for any insights!

[414N]

Hello Jack,

if I understood correctly, you started having these sudden hard-reboots only after installing the 3 HDDs into your system, right? Were you able to run heavy jobs (like the aforementioned "make -j12") without issues before?
If this is the case I agree that this sounds a lot like a PSU-related issue, likely caused by the added power load of the 3 HDDs.
By looking around, I gather that the power budget for your machine could be something like this

• CPU: between 135W and 368W (source)
• GPU: 185 W TDP (source);
• RAM: 5W per stick (source);
• HDDs: 15/20 W (source)

This does not take into account the power drawn from the motherboard itself or any other devices connected to the system. Also, keep in mind that the TDP values are not the maximum power draw that these devices can have, it can be fairly higher.
If the numbers for the CPU are to be relied upon, I think that with a 500W PSU you're already playing on the edge of its capacity without taking into account the GPU. Consider that a PSU (particularly if it's not gold or platinum rated) degrades over time and could become unable to provide its nominal capacity the older it gets. For this reason it's usually suggested to get a PSU that has a somewhat higher power capacity "buffer" than the one that the system is able to actually draw: the more high quality the PSU, the lower this buffer needs to be.
For this reason and given the aforementioned numbers, I'd suggest nothing lower than a 700+W PSU.

I’d prefer to stick with a 500-watt unit if possible just to keep the efficiency curve tight for idle usage (since this machine runs 24/7), but stability is obviously the priority.

I don't think that staying as close as possible to the actual maximum power draw that the system can exert will keep the efficiency curve optimal, I'd guess that's the opposite: keeping the PSU always working close to its max won't be so efficient, with the added risk that any sudden burst in power draw by any component could cause instability.
However, take this with a grain of salt, as it is something I just recall having read in the past.

Let's see if anyone else chimes in with other opinions too.