would be good to know what i7 it is, and if you have overclocked it

Can't remember all the details of the processor right now but I know i've never overclocked it.

Does your PC Freeze before this happens?
Is it heat?

I use to have this problem with my pc a year ago, due to it overheating and the cpu fan controller being plugged into the wrong socket.

More unlikely of a case, the same pc actually died recently, due to again the CPU, from this long period of time with my computer crashing, and me choosing not to take any action.

TLDR: Just make sure your computer isn't overheating.

As I wrote down all the specs above off the top of my head, I made a mistake about the CPU. It's actually an Intel Core i7 3770 3.4GHz. Sorry about that.

Nope, it doesn't freeze or do anything special; it just shuts down by itself (to clarify: when this happens the PC doesn't show the "turn off" screen as when you select it from the start menu, the case just powers off and the monitor goes to black).

I ran speedfan to check the temperatures, and they seem normal. I recently (like a couple of weeks more or less) cleaned the dust from inside the computer too.

1. memtest86. Because you never know.
2. Get a thermometer. Put it at the PSU's exhaust. Note the temperature when the PC crashes. Note it again once it's successfully booting again.
You can obviously do 1&2 at the same time.

This happened to me, I replaced my PSU and it stopped. Mileage may vary.

Just finished running it, didn't have any relevant issues (I guess?, I'm a noob with BIOS stuff)

Unfortunately I don't think I'll be able to measure it precisely, the only thermometer we have at home is a mercury one with a 40ºC limit :/

Ye, the PSU is still the most likely problem but I wanna be sure before spending money on a new one.

So you want to make sure it's the PSU, but are unwilling to do even the most basic test?

I mean if a 40°C thermometer isn't enough then either your room is insanely hot and the PSU is running outside of spec (which model exactly anyway?) or it's overheating.
You could find out by actually doing the test.

OK I'll try to do it, I was just saying that maybe those old thermometers weren't good enough to measure the exact temperature of the PSU.

I don't need an exact temperature. I mean if it's at 30°C it's probably not overheating, if the thermometer can't even show the temperature it probably is.

So I finally got these results:

• When it crashed (was working for about 2 hours): 37C
  
• When it loaded again (about 10 - 15 mins later): 32C

I got a same thing happening to me, but on an MSI laptop. After doing some research, turns out that multiple warnings of this type show up in the event viewer right before every crash - "The speed of processor 7 in group 0 is being limited by system firmware. The processor has been in this reduced performance state for 71 seconds since the last report."

However, not every warning like that is followed by a crash. I partially solved the issue by allowing maximum performance at all times when using my laptop, but it didn't solve it completely. I recommend checking for similar warnings (just use search on the taskbar and type 'event viewer', go to windows logs->system). This may narrow down the research you need to make. My laptop is not overheating either, that's why I started to check for other things.

Basically if it happens again, after a completely different time and the temps are the same then it's probably the PSU overheating.
You should be able to force it by running stress tests.

I looked for critical errors on the event viewer but didn't found anything relevant from when the crashes happened. I already have the max performance enabled too.

Ok, I'll monitor the temperatures from the next couples crashes to see if they are the same. Can you tell me how to run those stress tests tho? Thanks in advance.

Just any stress test. High GPU + decent CPU load would be Unigine Heaven/Valley.
Google them. Pretty self explanatory how to run them.

Alright, ran them both for about 30 mins and the temperature was slowly going up until getting capped at 76ºC. Did not get any crash tho. Should I maybe run the benchmarks for longer periods of time?

I hope not both at the same time.
Yes, run one a bit longer.
PSU temp.

If a stresstest can't force PSU overheating then it might be something else.

Like others said, got to watch the heat.

Oddly enough, yesterday I ran the Heaven one for almost an hour in the daytime and nothing happened (although the temperature went up to 77ºC at some points), but then the computer shut down in the evening while playing for 30 mins. The PSU temperature was even lower than the last time (33). Am I doing something wrong here? :/

Seems more like a driver crash then.
Have you updated the drivers recently?

I try to update them at least every month.

Tried updating/rolling back?
Also try logging the GPU temps to file with GPU-Z.

I rolled back the GPU driver to the last one, but it crashed anyway. I'll try to revert some others these next days, just in case.

Also, this is the log of the GPU-Z after 10 mins of playing:

• GPU Core Clock - 324.0MHz
  
• GPU Memory Clock - 162.0MHz
  
• GPU Temperature - 39.0ºC
  
• Fan Speed - 30%
  
• Fan Speed - 810RPM
  
• Memory Used - 133MB
  
• GPU Load - 2%
  
• Memory Controller Load - 3%
  
• Video Engine Load - 0%
  
• Bus Interface Load - 1%
  
• Power Consumption - 11.2% TDP
  
• PerfCap Reason - 16
  
• VDDC - 0.8750V

I need the file and preferably from after a crash.
Those are just normal idle stats.

So sorry about the late response, I had some other problems.

Anyway here is the file. It was taken about 5 minutes after the last crash.

No you don't understand. GPU-Z needs to be running during the crash. How am I supposed to figure out why it crashed without knowing what happened before the crash?
Whatever happened before the crash caused it, not what happened afterwards.

Logging 10 seconds of idling does nothing, this is the same as the screenshot before.

Just leave it running in the background.

OK I'm dumb. Here is the new log, crashed about 10 mins into playing TF2.

And it took you about 10 minutes before the PC properly restarted after that? Just making sure there's not too much missing.
Seems like your GPU is fine though.

Ok one last try.
You still have the problem that it won't restart for a while after a crash?
You know how to do the paperclip PSU test?
Wait for a crash, then try to start the PSU with the paperclip. If it doesn't then it really should be the PSU.

Sorry again for the delay.

The restarts after the crashes are pretty much random; sometimes the computer starts again without a problem and sometimes it takes some minutes for it to function again (although it tends to crash again after 30 mins anyway).

The first time I did the paperclip test the PSU didn't turn on, but then I checked all the pins and everything and tried again, and this time it worked. Oddly enough, while I was doing this my father saw it and asked me about it. I explained him the problem and told me that if we needed to change the PSU he could help me buy a new one, so we ended up buying it (Nox Urano VX 650W) on an internet store.

I installed it and it worked fine the first couple of days, but then the computer started crashing again. The product wasn't refurbished and arrived with the original packaging and all the other stuff too, so I'm not sure if the new one is malfunctioning or not.