|
Problem description: PC crashes after 10-15 minutes of either Crysis or the Unigine Heaven Benchmarks. Crashes as in screens go blank, then a couple of seconds later, sound stutters, then the PC shuts off entirely. If I hit the power button, it boots up fine (with the usual "Windows failed to shut down properly" screen), but the air coming from the PSU is like shoving my hand into a blast furnace. Attempted fixes: Updated video drivers, watched GPU/CPU temps while Heaven ran Recent changes: Last change was about 6 months ago - moved to an EVGA GeForce GTX 650 Ti SSC. Stock speeds for the card (which is factory OC'd), same with fan profiles, though I did use EVGA Precision to force the fan to 100% for the last benchmark. I did a full R&R of the latest GeForce drivers, from WHQL certified to the latest betas. e: Moved to the GTX 650 Ti SSC from a GT 520, no changes were made beyond that. 650 probably sucks down a bit more juice? Operating system: Windows 8.1 Pro w/Media Center addon System specs: Gigabyte Z68XP-UD3 (rev 1.0) with latest beta UEFI, 3 monitors (Acer 23" on DVI, Dell 21" on DVI, Toshiba 32" TV on HDMI from onboard video - optical audio output from onboard sound to Onkyo 5.1 receiver), Intel i5-2500k @ 4.2 GHz (was able to do 4.5 when on previous BIOS revisions), Cooler Master Hyper 212+ Cooler (PWM controlled with one 120mm fan), 8GB (4x4) Corsair 4GB sticks, 3x Hitachi SATA HDDs, 1x Samsung 120GB SSD, EVGA GTX 650 Ti SSC 2GB GDDR5, Corsair CX500 power supply, generic PCI fax/modem. The actual TV is turned on maybe once a week and runs off of the onboard video. All audio is via the Onkyo receiver via optical. If you care about networking, Verizon's latest FiOS gateway (gigabit from gateway to PC's onboard networking), one HDHomeRun Prime with a Verizon cablecard. LG SATA DVD-RW, generic 3.5" ("Power Up!" but identical to every  version) data card reader that handles CF, SD, etc. Wireless Logitech M510 (USB dongle), Logitech K120 wired keyboard. Onboard Intel HDMI is also enabled to handle the 32" Toshiba TV, but the TV is rarely turned on (and the PC doesn't even recognize it when it's turned off).Location: US I have Googled and read the FAQ: Mostly. I built this PC two and a half years ago; the only upgrades have been the OS (7 -> 8 -> 8.1), video card, and swapping out the main drive for a SSD. The only thing that brings it to its knees (especially to the point of a full power down) are the original Crysis and the Heaven Benchmark. I'm strongly suspecting my PSU just isn't up to the task - neither CPU or GPU temps get past low to mid 50s, even with all of the fans as low as they go. When it crashes, both of my main monitors go into power saving mode (I haven't tried it with monitor #3 yet, since it's a wall-mount TV that I rarely use). Num lock/caps lock/etc still respond for a bit, but ctrl-alt-del or ctrl-alt-shift do nothing. A few seconds after the screens go blank, the sound goes into a loop. Num/caps lock keys still respond, but anything to bring up, say, a task manager, or user menu, does nothing, so I'm forced to do a cold reboot (assuming the PC doesn't shut off beforfe then). In either case, I get the typical "hey rear end, you didn't shut down properly" menu. It really seems like anything super stressful on the video card is what triggers it - prior to Crysis, the most graphics-intense game I'd ever played was CounterStrike:GO. Which has me suspecting either a bad video card, or a power supply that's just barely adequate for the first few minutes. Also, I might get a BSOD once a year - every time so far it's been due to a HDD spinning down when the OS didn't expect it (easily fixed by disabling built-in power management) or a beta driver issue. I've watched CPU and GPU temps with CoreTemp and EVGA PrecisionX. The CPU gets up to the mid 50s during benchmarks (low 30s idling), GPU up to about 65C, on stock settings. If I get aggressive with the fan profiles, CPU sits around 40-45, GPU about 45C. My last test with MemTest86+ was perfect, though it's been a few months. The CPU and GPU heatsinks aren't clean enough to eat off of, but they get dusted every 1-2 months. One case fan is seized (unplugged at the moment), but it has 3 others (all 120mm). Case temp is generally 2-3C above ambient. HDD temps run around the same. Ideas? This is the first time I've ever had an issue with this PC. The sudden power off is telling me it's probably the PSU saying "gently caress you", but if I'm gonna replace it, it's getting a good (likely Corsair or Antec) modular 750W instead of something cheap. Just don't want to spend the money until I have to; I won't go non-modular for the next one, and it'll have to be a brand I like. I've lost a lot of Antecs over the years, and had an Enermax go up in flames once.  I have not had a way to monitor voltages while running yet, but while within the UEFI/BIOS setup, they're all rock solid and +/- 1.5% across the board.e: it has a bit of ricer LED lighting in it, but I think the LED strips combined draw about 1-2W @ 12V. randomidiot fucked around with this message at 11:55 on Jul 8, 2014 |
#
?
Jul 8, 2014 11:35
|
|
|
|
| # ? May 12, 2024 12:51 |
|
|
If the system's stuttering before it crashes, it could be leaving hints about what device is causing the issue in your event viewer. Checking that is free. You might also want to make sure your video card's power connectors are all plugged in properly, and that you didn't pull any of the wires out of the molex adapter when you removed your old card. If all that's squared away and you're confident it's not an overheating issue, then go with your first instinct and replace the power supply. That is quite the system you have there and things with moving parts (e.g., power supplies) do eventually fail.
|
|
#
?
Jul 8, 2014 15:37
|
|
|
The old card didn't have a molex connector - it was a budget card even back when I bought it. I just checked the power connector to the video card, it's secure. I had both EVGA PrecisionX and CoreTemp running on the 2nd monitor - temps were all well within acceptable ranges. I think the video card got up to about 55C, CPU was hanging out in the low 40s. And both monitors are going into power save mode about a second before the stuttering sound, which has me thinking whatever sound was still playing must have been buffered, or the video card is crashing before taking the rest of the system with it. I even forced the GPU and CPU fans to 100% at one point, which kept both of them just a bit above room temp under load - and still got crashes. Just checked the event viewer - only critical/warnings I'm finding around the time it crashes are the typical "The system has rebooted without cleanly shutting down first.". Some warnings about drivers for my multi card reader (SD/CF/etc), and I did find one "Display driver nvlddmkm stopped responding and has successfully recovered." - but that was while running video benchmarks (Heaven). Updating my video drivers to the latest betas resolved that. I'm on the fence between it being the PSU or video card now though - this morning, with the Heaven benchmark running, I tried increasing the GPU frequency by 100 MHz, and immediately got both screens to go blank and the system to power off a few seconds later. It's a factory overclocked card, so I'm wondering if I may need to RMA it. It still has 2 1/2 years left on the warranty, and I have a spare card to keep my system running while it gets shipped off, if it comes to that. I may underclock it a bit and see if it still crashes. randomidiot fucked around with this message at 08:26 on Jul 9, 2014 |
|
#
?
Jul 9, 2014 08:23
|
|