|
Problem description: I recently upgraded my Hackintosh from a Radeon 5770 to a GeForce 980Ti (specifically the EVGA GeForce GTX 980 Ti 6GB K|NGP|N). Why the 980 instead of a 1080? Because nVidia hasn't released drivers for the 1000-series cards for OSX yet. But I'm not even trying to deal with Mac stuff right now; this is all for Windows. The problem is that after running GPU-intensive activities (playing DOOM or running the FurMark stress-test) for a few minutes, both displays go blank, the sound starts stuttering, and the system crashes completely. DOOM lasts longer than FurMark (20-30 minutes vs. 3 minutes for FurMark), and I can do other stuff (watch YouTube videos; play Transcendence, an indie 2D space game) indefinitely. OpenHardwareMonitor says the GPU is peaking at 83C when running FurMark, and no other indicators (CPU temp, etc.) are high. When I reboot after a crash, after the Windows logo shows up during the boot prompt, instead of going to the login screen, the screen goes blank (first time it was blank black, second time it was blank blue), and I have to reboot again to get a usable system. I've rebooted pretty quickly after the crash; I haven't tried letting the computer sit to see if that results in a first-try-successful bootup. Attempted fixes: I replaced my old 700W PSU with a new EVGA 1050GS PSU, since NewEgg says I was pushing what the old PSU could supply. The new PSU should be total overkill for my system; it can put out 87A on the 12V rail. Drivers are fully up to date. Recent changes: Uh, I installed a new GPU and PSU.  -- Operating system: Windows 7 64-bit. System specs: Custom-built Hackintosh. Intel Core i7 3.2GHz, 24GB DDR3 RAM (16 of which is Corsair, I forget what the other 8 is). PSU and GPU as described above. Four hard drives: two 7200RPM platters, two SSDs. I can't remember what my motherboard is or how to check it without cracking open the case, which I'm heartily sick of doing today, so unless you think it's really relevant, I'll just leave it out. But the mobo is, like the rest of the computer, about 7 years old. Location: United States I have Googled and read the FAQ: Yes My hypothesis is that I got a bad card, but I figured I'd ask for a second opinion before I go to the work of RMAing the thing. EDIT: Windows' Event Viewer shows a bunch of "nvlddmkm" error-level events around about the time of the crash; that's probably related: event log posted:The description for Event ID 13 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer. And similarly for GPC 3, TPC 1, GPC 3, TPC 0, GPC 2, TPC 2, etc. etc. etc. Basically every combination of GPC [0-3] and TPC [0-2] TooMuchAbstraction fucked around with this message at 04:41 on May 11, 2017 |
#
?
May 11, 2017 04:24
|
|
|
|
| # ? Apr 25, 2024 14:37 |
|
|
Update: my display setup appears to make a difference. Here's the tests I've run: Default setup: * Old crappy display (HDMI port) * HDMI-capable flatscreen TV on an HMDI->DVI adapter (DVI port) The crappy display had been giving me trouble for awhile, so I tried removing it (using just the HDMI TV as my sole display), and I was able to run FurMark (GPU stress test) for 5 minutes, more than double my previous best. So I tried this setup: * HDMI-capable flatscreen TV on an HDMI cable (HDMI port) * Backup display on a DVI cable (DVI port) And it crashed during the stress test. So I unplugged the backup display and rebooted: * HDMI-capable flatscreen TV on an HDMI cable (HDMI port) And it crashed immediately, no stress test needed. So I unplugged the TV and plugged in the backup display: * Backup display on a DVI cable (DVI port) and FurMark ran for five minutes with no problems. The GPU has 1 HDMI port, 1 DVI port, and 4 mini-DisplayPort ports. I dug up a DVI->mDP adapter, and plugged the TV in via HDMI->DVI->mDP: * HDMI-capable flatscreen TV on an HDMI->DVI->mDP adapter chain (mDP port) * Backup display on a DVI cable (DVI port) and the system immediately crashed. Like, I heard the "Windows detected a new display" sound effect, then both displays went black. So, my hypothesis now is that this GPU for some reason just can't support more than one display, and that display has to be on the DVI port. Either that, or it refuses to speak HDMI to anything. Either way that seems like grounds for an RMA to me. Any counter-opinions?
|
|
#
?
May 13, 2017 02:26
|
|
|
TooMuchAbstraction posted:Either way that seems like grounds for an RMA to me. Any counter-opinions? Maybe. Make sure you're using the latest BIOS for the motherboard.
|
|
#
?
May 13, 2017 20:33
|
|
|
Zogo posted:Maybe. Make sure you're using the latest BIOS for the motherboard. I'm curious as to your hypothesis; what would be the explanation for the failures I'm seeing? I'll also note that the old card (the Radeon 5770) was able to use its HDMI port just fine.
|
|
#
?
May 14, 2017 15:25
|
|
|
TooMuchAbstraction posted:I'm curious as to your hypothesis; what would be the explanation for the failures I'm seeing? Newer graphics cards are sometimes incompatible with older motherboards. Sometimes a BIOS update has been rolled out to resolve the issue. You may be able to find the motherboard info out without opening case: http://www.wikihow.com/Identify-the-Motherboard
|
|
#
?
May 15, 2017 00:44
|
|
|
I see, thanks for the explanation. I actually happen to still have the old mobo box. It's a Gigabyte X58A-UD3R, and the BIOS dates from 2012/01/06, which is the most recent version per this site. If that's causing incompatibility with the GPU, then my only options would be either to return the GPU or upgrade the mobo, and I don't really fancy dealing with a mobo upgrade just now. Any ideas on how I could tell if it was a mobo incompatibility as opposed to a GPU fault? EDIT: This page says that my mobo should be compatible with the 1000-series Geforces, so I assume I'd be fine with the 900-series.
|
|
#
?
May 15, 2017 02:33
|
|
|
TooMuchAbstraction posted:Any ideas on how I could tell if it was a mobo incompatibility as opposed to a GPU fault? Not any simple ones that I know about. A lot of times with the GPU/motherboard incompatibility the computer won't even POST. Also, you've tried other HDMI cables right?
|
|
#
?
May 15, 2017 05:20
|
|
|
Zogo posted:Not any simple ones that I know about. A lot of times with the GPU/motherboard incompatibility the computer won't even POST. It's a bit academic now; the old card's on its way back to Amazon and a new one should arrive tomorrow. But I think I did? Certainly I tried using the same cable that worked with the Radeon 5770, and I also tried an HDMI->DVI cable. Which worked fine when plugged into the DVI port, and failed when a mini-DisplayPort adapter was used to plug it into one of the mDP ports.
|
|
#
?
May 15, 2017 15:00
|
|
|
|
| # ? Apr 25, 2024 14:37 |
|
|
The first new card got lost in transit, but the second new card finally arrived, and seems to be working just fine. Slightly different model of 980ti, but I did FurMark for 5 minutes while using DVI and HDMI ports and the system's still running.
|
|
#
?
May 29, 2017 22:18
|
|