I installed a GEForce GTX 1050 and now my computer reboots a few seconds after the "Loading Initial Ramdisk" message.

My PC specs

Power supply DX-520WPS (520W)

I'm on Debian Buster/Sid, but this also happens when booting to Slackware-current (post 14-2), and FC 23.

A few months ago, when I was on Debian Stretch, I bought a previously owned GeForce GTX 750TI and had a similar problem. Those reboots seemed more random though. Sometimes I could boot to my desktop but then I'd reboot a minute later. I thought it was a bad card and returned it.

I've installed the drivers using the nvidia-driver (390.87) package from the Debian repo.

I tried flashing my BIOS last night with rev 0502 of CG5275.

Via /etc/default/grub, for my kernel line, I've added things like "nomodeset", "gfxpayload=text", "video=vesaff", "enable_mtrr_cleanup", and "mtrr_spare_reg_nr=2". So far, no effect.

While searching for solutions, I found results that indicated a grub bug may be the problem. I tried booting with lilo.. same result.

To my initrd image, I tried adding these modules:

I've replaced my CMOS battery.

I have PEG/IGD set in my BIOS.

Sometimes my computer reboots only once before proceeding to finish. Today it was 6 times.

Once I tried booting to Windows, and that worked. But only tried once so will need to do more "booting experiments" to see if there are any differences.

I realized this wasn't too clear in my first post, but after a reboot (or more), my PC does succeed to boot to my desktop, and then runs fine. The new card works great!

Some new info: One successful boot to Windows 10, and then one resulted in a reboot at the splash screen.

I removed the "quiet" command from my grub configuration. I've attached a screenshot that shows what appears in the console immediately before reboot.

After all the different things I've tried and the results I've observed, pretty sure now that this is just some incompatibility between the GPU and my motherboard, BIOS, or something funky with the PCIe slot. I got the computer used 3 years ago and have no idea if any GFX card *ever* worked in it (other than IGD).

Attached Thumbnails

 

That power supply might have a marginal component or two in it. I would swap with another to see if it makes a difference, after opening it up to see if it has any obviously leaking or swollen capacitors. All their tops must be flat. Also note the brand names that you can make out. Check out the pix on badcaps.net if you have any question about what you see inside. They have lists of cap manufacturers that are reliable, and those that are not. If the names are on the bad list, the PS probably is your problem eventually even if not now.

You probably have have no need for all the power the DX claims to provide. Most PCs can run on 300W or less, if that's what the PS can really provide reliably.

If you haven't run memtest86 for several hours or more, that's the next thing to try. Overnight should be enough, but sometimes it takes even longer to find that single bad bit or chip that works fine most of the time.

1 members found this post helpful.

@mmazda, thanks. I don't have an extra power supply to try swapping. I just ordered a new computer.

Most PC power supplies, including yours, are generic and easily swapped, and sold if not a solution to the problem and money or space consumed by a spare is an issue.

oh, I just meant that instead of buying and trying a new (or used) power supply, I'd rather just upgrade my desktop (esp since it's a decade old). I ordered a previously owned desktop with an i5 6th gen chip. I'll try the card in that and let ya know.

If you don't want to keep the old one, maybe my nephew can pick it up and bring it here. He spends several months a year somewhere around Minneapolis. I never had a socket 1156 generation, (or anything Asus branded) and could use it as a troubleshooting tool.

I'll keep that in mind and PM you if I decide to get rid of it. Right now I'm leaning on keeping it for a while, but that may change. It can still compile stuff at a decent speed, and I may use it as a "primary-secondary"

I got my new computer a week ago and put the geforce card into it. The first time I powered it up, it rebooted. I suspect that was from the BIOS auto-updating, because it hasn't rebooted since.

I let my old PC sit for a week, and yesterday did a fresh install of Debian stable, spent about an hour setting up syncthing for the first time and then powered it off after a few hours.

Today 4 reboots within an hour, and keep in mind that's using the Integrated VGA port that never did that before.

So based on what mrmazda suggested, I suspect all those reboots (from after I put the geforce card in) caused some problems with my power supply, or if it was marginal, sped up the process of making it "bad".

I'll check the caps as mrmazda suggested and probably replace the PS, and then report back here.

The innards of my power supply:

Attached Thumbnails

     

Far too much accumulated crud to tell if any caps are obviously bad. It might be OK simply by cleaning with compressed air outdoors wearing a dust mask and coveralls. Worst PC PS innards I've ever seen. Yuck!

The rest of that PC almost certainly needs cleaning too if you didn't already when you installed the GTX 1050.

1 members found this post helpful.

Really? That's the worst you've seen? This is pretty tame from some of the systems I've worked on. The heatsink only has a bit of build up on it, it isn't completely covered.

My guess is that it is just older hardware that is starting to go out. Plus it's unlikely a heat problem would occur that quickly after bootup. Even misplaced heatsinks on CPUs will allow systems to boot up and function for a bit of time before there's too much heat.

1 members found this post helpful.

The rest of the PC actually looks pretty good.

I already threw out that power supply. Next time I'll just try cleaning it with compressed air. That would have been better in hindsight.

So... now with the new power supply in (same make and model), noticed only a couple reboots. After 20 or 30 minutes it freezes if the video output isn't active. What I mean by that is that it's hooked up to a kvm switch on "port" 2. My main (new) pc is on port 1. If I turn on the PC on that's on port 2, it's fine for about 30 minutes, then it freezes.

I did some testing and changed some settings. One thing that worked 100% is adding "acpi=off" to the kernel command line.

Another thing that worked, before turning off acpi, was periodically switching back to video on port 2.

The freeze thing actually happened with the first PS, I just forgot to mention it. So.. seems the reason I was able to do a fresh install of Debian and syncthing (as mentioned above) was because the video was active the whole time.

Now no reboots the last couple days (but they *did* happen after the new PS was put in), just the freezes which have reproduced consistently. So basically this new problem is unrelated to my OP.

I could mark this thread as solved because my new PC doesn't reboot with the new GPU in it. Don't really know what caused the original problem still but could probably determine for sure if the PS were the issue if I put it in the old PC now that it has a new power supply.

EDIT: Looking more closely, I do see some build-up over the CPU heat-sink.

Attached Thumbnails

   

To me it looks like a lot more than just "some".

1 members found this post helpful.

This is all what I've done in chronological order, leaving acpi enabled:

Btw, when I mentioned "freeze" above and below, I mean that I couldn't ping or ssh in, and pressing the power button didn't shut the computer down (unless held for 5 seconds), and when I switch back to port 2, the display is blank and I can't reactivate it by pressing CTRL, or switch to a terminal with ctrl-alt-f1

Plugged into port 2 of the kvm, port 1 being the active port...

1. Booted to Slackware-current

result: reboot after ~20 minutes, then freeze after ~20 minutes

2. Cleaned cpu heat sink and fan. (see attached pic below)

3. Plugged in 4 pin connection to motherboard that I apparently had forgotten about (see attached pic below)

4. Booted to Debian stable

Result: freeze after ~20 minutes.

Attached Thumbnails