I actually couldn't get any help from OCN.net of all places for this, perhaps I can here.
I'll try to summarize for the tl;dr generation.
Have 780 Ti Classified, 2500k, 5 year old mobo.
Temps are fine, memtest passed, hdd's are fine.
GPU/CPU are not overclocked and the GPU has never been overclocked.
I was gone for a bit yesterday, then launched Rust. Right as what should be the first frame of the map after loading onto the server displayed, it froze, Screen was garbled like those block puzzles with one free space. Subsequent reboots resulted in Windows half booting or not full booting and nothing working on the screen, display driver crashing, or the system booting normally 20 minutes later and even some games working, but 1 in 90% of the screen goes black and the top 10% has garbled horizontal lines.
PCI-E slot has been pulled out a bit due to the weight of the GPU, GPU heatpipe doesn't constact some mosfets or PWM's on the edge (small cylinder things that looked like pwms), if that was an issue it would have mattered on day 1, not 2 years later as I noticed this when the card arrived.
GPU is still under the 3 year warranty and I'm the original owner but never registered, still have Newegg proof I ordered it with matching names and and show my ID (Not sure how EVGA requires you to proof it)
If this was somehow a PCI-E slot issue it would simply work or not work, I even fail to see how this is a GPU issue.
The only thing that makes sense is a memory issue and only when it's near full it may try to use corrupt memory and then fails ?
This isn't a PSU issue, I don't think. Furmark doesn't behave like that, if it wasn't drawing enough power it would instantly restart and I wouldn't see what I see.
I ran Furmark initially and it passed, GPU never got above 73c (780 Ti Class is designed to run at 83 on max load without going above 70% fan speed due to sound), all other temps are great. I ran Furmark again and it instantly froze and crashed.
>>58560011
but 1 = but one minute later
>>58560011
Download afterburner or whatever you want to use and underclock the card, see if that fixes it. If it does, RMA and complain since your GPU is at least kinda fucked.
>>58560076
I'm already using AB, I monitor it regularly.
I wasn't aware you could underclock with it.
What exactly would I change ?
>>58560098
Turn down the core clock down by 100 or 200MHz, boot up rust and see if it still fucks up.
>>58560122
I assume I should not check 'apply oc at system startup' in case it would get caught in a cycle of rebooting ? I know safemode could break the cycle.
>>58560132
Yeah, don't worry about that. If underclocking helps, RMA the card, you don't need any of that bullshit. Bonus too, someone like EVGA would definitely bump you up to a current gen card after an RMA.
>>58560166
I underclocked it by 105 MHz, can't go any lower.
Rust ran for 15 seconds then the system froze. GPU never went over 71C, no CPU core was above 42C.
I had to do a hard reset.
As for a new GPU, that would be amazing.
>>58560166
I seriously got more help from a first post on 4chan than from Overclock.net.....
Thanks so far btw.
>>58560166
Got the serial number, going to start the RMA process, not sure if I even can since I don't have that 2 year old invoice.
good luck with the RMA OP
>>58560685
Thanks
In chat with EVGA, confirmed to be under warranty, expires this July too, wow.
So what's with this upgrade thing ? That wasn't my goal at all, I just want a working GPU.
RMA approved guys, yay.
What's the chance of getting back a used 780 Ti ?
>73c
Ain't bad. My XFX r9 390 gets to 90c when I play elite dangerous
>>58563108
780 Ti Classified was designed to hover at 83c on full load while not going above 70% fan speed to drastically limit the noise, it's worked like that too.
I just have no effing idea what happened.
>>58563108
90c
Jesus fuck
I haven't seen a GPU that hot since my BFG 6800 Ultra OC from like 2004 when It had so much dust in it, it was basically like a blanket was interwoven in the heatsink.
I learned my lesson then.