Video output dies while gaming

[guildwarrior99]

Alright, before I call Digital Storm again, I figured I would turn here. I have read multiple threads about this issue, and I have a feeling I know the answer, but I figured I'd explain the issue from its origin and list my specs just in case anyone here might see something I (or Digital Storm) didn't.

I'll try to get through the background as quickly as possible. When I first received my computer from Digital Storm (the Vanquish 5 "Better" configuration), I immediately began experiencing issues with the video cutting out while gaming. I called them up and they had me physically remove and slot back in the card. This worked for weeks, as I was able to play 1080p60 games as well as VR games without a hitch.

Weeks later, I began seeing artifacts on the screen during gaming sessions and the computer would eventually emit a terrible noise (from the monitor, not the machine itself) and would shut off. Never on Hearthstone or while browsing the internet, but any time I'd try to play a game with a sizable footprint—an MMO like Riders of Icarus or World of Warcraft. Removing and replacing the graphics card seemed to help temporarily, but inevitably, the artifacts would reappear and the computer would shut down.

After receiving a new GPU from Digital Storm (which appeared to have issues from the minute I plugged it in), I sent the system back for repairs. I just got the computer back this week, started playing Riders of Icarus again, and 15 minutes later, the video output stopped again. No whirring fans, no shut downs—I just couldn't see anything and had to restart my system to get a signal again.

I've tried multiple monitors and multiple HDMI cords. I'm still under warranty, but before I pay for shipping again, I wanted to see if anyone here had some ideas that might not require me to replace the PSU (or the GPU for the fourth time). Thanks in advance. Here are the specs:

Processor: Intel Core i5 6500 3.2GHz
Motherboard: ASUS/GIGABYTE (Intel H110 Chipset)
System Memory: 8GB DDR4 2133MHz
Power Supply: 600W EVGA/Corsair
Optical Drive: DVD-R/RW/CD-R/RW (DVD Writer 24x / CD-Writer 48x)
Storage Set 1: 1x SSD (240GB Digital Storm Certified Performance Series)
Storage Set 2: 1x Storage (1TB Seagate / Western Digital / Toshiba)
Internet Access: High Speed Network Port (Supports High-Speed Cable / DSL / Network Connections)
Graphics Card: NVIDIA GeForce GTX 970 4GB
Sound Card: Integrated Motherboard Audio
Extreme Cooling: High Performance Cooler with Large Fan and Copper Pipes
Cable Management: Premium Cable Management (Strategically Routed & Organized for Airflow)
Chassis Fans: Standard Factory Chassis Fans
CPU Boost: Standard Intel Turbo Boost 2.0 Automatic Overclocking
Windows OS: Microsoft Windows 10 Home (64-Bit Edition)

 

[amtseung]

That doesn't sound like a dead GPU, that sounds like a dead/dying motherboard. Or heat, you might be overheating, and when the graphics card hits that critical temperature, it'll just quit displaying.

I have a hunch, that the motherboard's PCIe slot that the graphics card occupied damaged itself in transit, which is why Digital Storm themselves couldn't figure out the problem. Plug the display cable into the back of the motherboard and remove the graphics card from the system. Do you still randomly lose display signal?

 

[guildwarrior99]

That doesn't sound like a dead GPU, that sounds like a dead/dying motherboard. Or heat, you might be overheating, and when the graphics card hits that critical temperature, it'll just quit displaying.

I have a hunch, that the motherboard's PCIe slot that the graphics card occupied damaged itself in transit, which is why Digital Storm themselves couldn't figure out the problem. Plug the display cable into the back of the motherboard and remove the graphics card from the system. Do you still randomly lose display signal?

It might be hard to replicate using the motherboard's onboard graphics, because it only seems to happen when I'm running a graphically intense game, which I assume wouldn't even load with whatever chipset is inside the motherboard. But when I have plugged the HDMI cable into the motherboard rather than the GPU in the past (just because I didn't know what to do and I'm not much of an expert or even an intermediate user when it comes to computer hardware), I don't remember having the same issues. I can still try again, with the GPU removed, but I'm not sure how to test it since the computer can run for hours on end without experiencing any issues until I try to run something that pushes it hard.

 

[amtseung]

Download Unigine Heaven or Valley. Either will do. Open it and just click "run". The settings don't really matter. This will stress just the GPU, to isolate its load so we can see what's going on.

Plug the graphics card back in and everything and run the Unigine program (whichever one you downloaded). Don't benchmark it, just let it loop until the display fails. All the meanwhile, keep an eye on your GPU temperatures. Unigine should have an option in the in-benchmark settings to display temperature, clock speeds, etc.

Now loop the same Unigine program with the GPU removed from the system, running it off the integrated CPU graphics, plugged into the motherboard instead of the graphics card. Will the display signal fail then?

 

[guildwarrior99]

Download Unigine Heaven or Valley. Either will do. Open it and just click "run". The settings don't really matter. This will stress just the GPU, to isolate its load so we can see what's going on.

Plug the graphics card back in and everything and run the Unigine program (whichever one you downloaded). Don't benchmark it, just let it loop until the display fails. All the meanwhile, keep an eye on your GPU temperatures. Unigine should have an option in the in-benchmark settings to display temperature, clock speeds, etc.

Now loop the same Unigine program with the GPU removed from the system, running it off the integrated CPU graphics, plugged into the motherboard instead of the graphics card. Will the display signal fail then?

With GPU plugged in: The temperature kept climbing from the minute I started the Heaven benchmark. Started in the 50s, went to 60s, 70s, 80s and then as it approached 90C, the FPS collapsed down into the 20s. Hit 96C before the display killed itself again. Clock speeds seemed to stay consistent throughout? Will try turning everything off and redoing test without the GPU now.

 

[amtseung]

No need to rerun the test. Your GPU is overheating. Anything over 80C is worrying. Now we need to figure out why.

Are the fans spinning faster as the GPU gets warmer? Is the GPU cooler properly screwed securely together?

 

[guildwarrior99]

Yeah, I knew 96C was way, way too hot based on what I've read around here. (And integrated CPU test cycled through without a sweat, just FYI). As for the fans, I'll run the test one more time to make sure, but the computer definitely remained almost entirely silent the first time through, even in the 90s (if that matters). Everything seems to be screwed in properly. Maybe an ever-so-slight wobble to the Zalman unit if I press on it, but not like it's about to fall out or anything. By the way, here's an image of my system (if this works): http://imgur.com/nUfnt8p [fixed link]

 

[guildwarrior99]

No need to rerun the test. Your GPU is overheating. Anything over 80C is worrying. Now we need to figure out why.

Are the fans spinning faster as the GPU gets warmer? Is the GPU cooler properly screwed securely together?

The fans (both the one at the back and the one above the GPU) started running fairly quickly as soon as I started the test and seemed to continue spinning at the same speed throughout the test. It approached 90C again.

 

[amtseung]

Download EVGA Precision X (or precision OC or whatever they call it now), find the fan speed slider thing, and max it out, and run heaven again with the side panel removed. If the temps are still steadily climbing, try applying some new thermal compound to the GPU. Here's a good old Jayztwocents video to walk you through it.

Otherwise, it might be time to complain to Digital Storm to see if they can't replace the graphics card.

 

[guildwarrior99]

Download EVGA Precision X (or precision OC or whatever they call it now), find the fan speed slider thing, and max it out, and run heaven again with the side panel removed. If the temps are still steadily climbing, try applying some new thermal compound to the GPU. Here's a good old Jayztwocents video to walk you through it.

Otherwise, it might be time to complain to Digital Storm to see if they can't replace the graphics card.

Tried moving the fan slider up to high and took the side panel off. Fans were whirring (I think about the same as they would have been anyway), but the temp was increasing just as quickly. And as a fun added wrinkle, this is the third GPU that I've had in this system. Virtually the same issue every time. Started with an EVGA GTX 970, they sent me an ASUS card that seemed to be busted and now I have a third card (also an ASUS), and the issue is still cropping up.

Sorry for the long delay in response, but really appreciate your help so far.

 

[amtseung]

If the same thing has happened with three separate graphics cards, something else is wrong. I wonder if it's a stray incorrect voltage setting or a bad power supply. Blower coolers don't suffer from bad temps due to bad case airflow anywhere near as badly as recirculating coolers.

I'm stumped now.

 

[guildwarrior99]

If the same thing has happened with three separate graphics cards, something else is wrong. I wonder if it's a stray incorrect voltage setting or a bad power supply. Blower coolers don't suffer from bad temps due to bad case airflow anywhere near as badly as recirculating coolers.

I'm stumped now.

I might be grasping at straws here, but is there a chance it could be something in the BIOS settings? Or could any other component in the system possibly be causing the GPU to overheat? Providing it's some hyper-specific hardware issue, I will likely need to send it back to Digital Storm anyway, but the fact that they seem to have not even run a graphics benchmark when testing it doesn't leave me with much confidence in their ability to fix this.

 

[amtseung]

You should definitely link them this thread, so they can see what's going wrong.

The bios you can access by pressing delete or f2 or whatever is the motherboard bios, which controls, very specifically, motherboard-related things. Graphics cards have their own bios. You can't exactly access it, per se. Graphics card manufacturers tend to really hate it when we do mess with their bioses.

Now that I've slept on it, the last thing I can think of is that your motherboard has a faulty PCIe slot and is pushing too much power through the slot, or your power supply is faulty. An increase in heat can is always caused by an increase in power consumption, so something somewhere is feeding your graphics card too much power under load, which isolates the motherboard or power supply.

 

[guildwarrior99]

You should definitely link them this thread, so they can see what's going wrong.

The bios you can access by pressing delete or f2 or whatever is the motherboard bios, which controls, very specifically, motherboard-related things. Graphics cards have their own bios. You can't exactly access it, per se. Graphics card manufacturers tend to really hate it when we do mess with their bioses.

Now that I've slept on it, the last thing I can think of is that your motherboard has a faulty PCIe slot and is pushing too much power through the slot, or your power supply is faulty. An increase in heat can is always caused by an increase in power consumption, so something somewhere is feeding your graphics card too much power under load, which isolates the motherboard or power supply.

I'll be sure to let them know what we've gone over here. Thanks again for your help.

 

[amtseung]

Sorry I can't help any more. I want to, and this issue has certainly got my brain churning, but my limited technical knowledge is preventing me from doing so.

I spoke with a friend of mine who works for Nvidia (he does memory control design architecture whatever stuff, so I assume he's very well versed in things) about this, and I showed him this thread. He was equally stumped by how you got the same result with three different graphics cards. The only thing he could tell me with any certainty was that chances are, all three graphics cards were actually perfectly fine, and something else is broken here.

 

[guildwarrior99]

Sorry I can't help any more. I want to, and this issue has certainly got my brain churning, but my limited technical knowledge is preventing me from doing so.

I spoke with a friend of mine who works for Nvidia (he does memory control design architecture whatever stuff, so I assume he's very well versed in things) about this, and I showed him this thread. He was equally stumped by how you got the same result with three different graphics cards. The only thing he could tell me with any certainty was that chances are, all three graphics cards were actually perfectly fine, and something else is broken here.

So, this isn't really a new wrinkle, but rather me slowly building upon my extremely limited knowledge of PC hardware. It turns out that the GPU fan in the ASUS GTX 970 (the third and latest card) never starts spinning. From what I've read, this is supposed to start at around 65C, but mine never starts, even as it hits the 70s, 80s, 90s. I realize that this could be the card itself, but as we discussed, three straight faulty GPUs would be an insane coincidence.

Is there something on the software side I can try here? Maybe in the BIOS menu or something? I've tried downloading GPU Tweak (the ASUS version of Precision XOC), and no matter what I do to the GPU fan, it never starts spinning.

On the other hand, could this be going back to the original diagnosis, that something is wrong with the way that the GPU is connecting to the motherboard? Or some issue with another piece of hardware?

 

[amtseung]

There should be a fan plug going from the cooler portion of the graphics card to the PCB itself, usually on the end near the power plug for single fans, and on opposite corners for dual fan cards.

From the factory, the fans should be programmed to follow a certain fan curve, dictated in the card's own bios. The only thing that would prevent the fan from spinning is either the plug has fallen out (happens surprisingly often when the things get shipped across the country, picture of a fan header here), or there is a physical obstruction within your case (a stray wire, plug, or zip tie) that is preventing it from spinning.

If this only happened with one card, there would be many more possibilities, but since it's happened with three from multiple manufacturers, and the only commonality is that all of them were shipped from digital storm (does that make it overseas?), that leaves us with those two potential hardware failures.

Unless you have conflicting control softwares installed, and the one that takes priority tells the fan to never turn. None of the control softwares can make the fan spin? Even when you go manual fan percentage and crank it to 100%?