I had some strange issues today with my pc. Every time I loaded a game it would run for about two minutes and then my PC would shut down with the graphics cards fans blowing at full tilt. I had a windows update the night before so I thought it had something to do with that. I uninstalled the graphics drivers and did a clean reinstall. But again every time I tried to play a game the PC would shut down. I started to think it was a hardware problem. So I tested the graphics cards individually, pulled out the ram an re-seated the dims, you name it. Then I started to suspect the PSU. I checked the air filter but that was clean so I checked if all the cables were in place. I found a charred PCIE cable that was shorting. I think it got twisted when I turned the PSU upside down a couple of weeks ago. Luckily I had a spare cable and all seems to work just fine now.

Looks fine to me.

yeah.. that pix is sooo hot.. hrmm yeah..

I once forgot to plug the fan cable back in. But the PC just shut off. They don't catch on fire, lol.

What PSU do...er, did you have

Yeah I had something like that happen. Had an inexplicable smell coming out of my PC. Turns out it was the smell of melting plastic. One of the connection from my GPU was doing something weird and melting the connecting to the mobo. Good times.

@hakunin: A corsair 850w Silent Pro M2 (bronze rated). I'm still using it but just not the damaged socket. I will be replacing it ASAP. I ordered a Corsair RM850x (gold rated).

@Bdead

Recent AMD cards had an issue, where they drew more than the spec'ed 75 W from the PCI-E, though in this case it's the extra power cable that got melted. Video cards draw rediculous amounts of power and also they aren't very well tuned on factory settings.

900 series GTX and earlier Nvidia cards also had issues where they ran at full clock/full blast when in dual monitor mode or high resolutions, even when idle and no games running.

You should look into downclocking your card a little or at least tweaking voltages down a bit, at the very least you should take note of your consumption.

yep looks pretty good.. to me either.

that's danger illustrated

@eurobum: I have 2 1080ti's and at full load my PC draws about 700W. I'm pretty sure it was the cable that got damaged and shorted. The PSU has very good protection so I guess it's fine but I'll be replacing it nevertheless.

@bdead: Wow that is a rather exuberant setup, which works close to the amperage limits of all the interfaces. Tomshardware hired a pretty good reviewer, who has some experimental chops. They do detailed analysis of consumption, which is very illuminating. You'll see massive spikes in their graphs a card with a 250W TDP can go well beyond 300 W briefly. At 20-30 A currents it takes just a bad connection or a slightly thinner cable to cause runaway cable heating and burn, even though the current is spit between all of those connectors. Every 6 pin is 2x8 Ampere max or something ...

SLI setups are also a different beast, and the first card has to work harder than the other. They can only use the GDDR memory of 1 card, AFAIK.

I don't ever use setups like that or would ever recommend them, so I have no experience. But it always pays off to tweak stuff. Perhaps clock down the GPU and perhaps clock up the memory? Running SLI with a factory OC is completely unnecessary and these big cards are always bandwidth starved.

Run some benchmarks before and after, make some screenshots, take notes. Science!

@bdead: Have the RM850 from a couple years back and works like a charm (though I've only ever had a single GPU).

Was there no smell?!