Bunch of Hardware Failures, centered around 7970/280X

[redsox0914]

My computer died about two weekends ago. Lights on motherboard and CPU cooler (Antec 920) are on, but the power button doesn't work at all. I assume it's either the motherboard (ASUS P9X79 PRO) or the processor (3930K). I'm not sure which one would be the culprit here (I'm thinking probably CPU, but not sure).

I had a backup system (MSI 890FXA-GD70 board, 1090T processor) so I started using this backup with my 7970 and two hard drives. Everything works fine until this past weekend. Display suddenly goes out (black screen, no-signal indicators on monitor) and there's no way to get it working. I swap back to my ancient 3870 in the meantime so I have something to work on, and order an open box 280X from Newegg.

The 280X arrives today. Swap it with the 3870 and everything works fine so far. Reinstall the video card drivers, no issues yet. Open up a game, and I get black screen/no signal again. The computer turns back on, but will not run the game. At this point I decide to try swapping the backup PSU (900W Rocketfish PSU from Best Buy, lol) with my main one (1000W Rosewill PSU).

I get a bit greedy and try hooking up the 7970 as well with the 280X (hoping that it was just the display adapter part of the 7970 that broke, or that the 7970 is somehow fine, and that the chip itself can still crossfire). Computer will not stay on for more than 5 minutes now, and this includes after I've removed the 7970. Even if I'm in BIOS or booting off the DVD drive to try to reinstall Windows the computer will not stay on.

I'm back on the 3870 now, which is the only thing I have at the moment that will run without the computer shutting down.

A bit of history on both systems. My older backup system has run crossfire 5850s for multiple years. Newer system has also run the 5850s and had a single 7970 running since its release until 2 weeks ago.

Any ideas which parts might be bad/dead, or what else I could test to get more information? Thanks in advance.

 

[lavaheadache]

I would start by investigating your psu's a little bit.

 

[redsox0914]

Two questions about this:

How can I test them, and what are the chances that I'll have had two go bad within two weeks of each other like this?

 

[lavaheadache]

lets leave odds out of this. Time and time again, things happen that beat all odds. You can test them with a multimeter if you know what you are doing but it's not something I feel comfortable telling you how to do over the internet

 

[Termie]

A failing video card will typically not cause a computer to simply shut off. When you say the computer will not stay on, do you mean you're getting bluescreens, or do you mean the computer simply shuts off?

And does the HD 3870 work in your original Intel system, which appears from your post to have been the first one to fail?

And was the 7970 in the Intel system to start with? You never said in your post.

By the way, I wouldn't have touched an open box R9 280X...

 

[redsox0914]

It usually just shuts off or the signal goes blank after a few minutes, including when I'm not in Windows. Sometimes I'll still be in BIOS and the signal will disappear. I've also had it happen while loading the Windows 7 install DVD. No bluescreens.

I haven't tried the 3870 in the Intel system. I didn't even think about trying that and just threw the 7970 into the backup system.

7970 was in the Intel system to start with, yeah.

As far as the open box 280X, yeah I may have gotten desperate there. I figured it was probably a card someone couldn't overclock well and sent back, and I have little interest in overclocking in the meantime.

 

[TeknoBug]

Sounds like the PSU or motherboard is going, a video card doesn't cause that sort of stuff.

 

[BrightCandle]

Sounds like the PSU is failing. If hardware like the GPU was doing it you would expect a blue screen or the machine to lockup but remain on. The turning off without warning is almost certainly the PSU cutting out.

 

[Arkadrel]

Two questions about this:

How can I test them, and what are the chances that I'll have had two go bad within two weeks of each other like this?

Insanely small.

Its probably something else.... Could be the PSU.

 

[Despoiler]

I bought a PSU tester to make troubleshooting stuff like this easier. The below one is the one I got, but several companies make them. Very useful IMO.

http://store.antec.com/Product/accessories-other/digital-pc-power-tester/0-761345-77204-4.aspx

 

[amenx]

This is why I would never be without a second PC. Simply to test things from the other if there are suspect parts, driver/software conflicts or other system glitches. It would then be very easy to rule things out. Not to mention not having to wait for replacement parts if or when required.

 

[digitaldurandal]

A failing video card will typically not cause a computer to simply shut off.

It sure can, not sure why you are stating it won't.

I have a 6570 with a failed fan and the computer shuts off completely due to overheating.

 

[Termie]

It sure can, not sure why you are stating it won't.

I have a 6570 with a failed fan and the computer shuts off completely due to overheating.

A system's response to an overheating video card would likely differ from a system's response to a video card that has failed. Distinct scenarios.

It usually just shuts off or the signal goes blank after a few minutes, including when I'm not in Windows. Sometimes I'll still be in BIOS and the signal will disappear. I've also had it happen while loading the Windows 7 install DVD. No bluescreens.

I haven't tried the 3870 in the Intel system. I didn't even think about trying that and just threw the 7970 into the backup system.

7970 was in the Intel system to start with, yeah.

As far as the open box 280X, yeah I may have gotten desperate there. I figured it was probably a card someone couldn't overclock well and sent back, and I have little interest in overclocking in the meantime.

So, OP, I suggest you troubleshoot this in a very disciplined manner by looking at all the facts you have in front of you:

(1) An Intel X79 system with an HD7970 and Rosewill 1000W power supply "died" a few weeks ago (you should specify whether the symptoms were identical to what you found later on a backup system, by the way).
(2) The HD7970 caused random system shutdowns on an AMD-based system using a generic power supply, whether in the BIOS, Windows, or a game.
(3) An open box R9 280X purchased for testing exhibited the exact same (confirm this) symptoms.
(4) Using the Rosewill 1000W power supply in the AMD-based system did not alleviate the problems.
(5) An HD3870 video card works fine in the AMD-based system, with the generic power supply. Has not yet been tested in the Intel system.

There are three variables we're working with here: the video card, the power supply, and the motherboard. All could fail on their own, but it's unlikely you have two bad video cards, two bad power supplies, and two bad motherboards. I would start by trying the HD3870 in the Intel system with the Rosewill PSU, and assuming it works, you likely have two bad video cards. The one unknown there is whether it's in fact the power supply unable to handle the power draw of the HD7970 even at idle. I don't really trust the generic 900W power supply as proof that it's not a power supply issue. But start with the 3870 in the Intel system and then work from there.

By the way, an open box R9 280X most likely was returned by someone who mined with it and then lost interest. That could have put a lot of wear on it.

 

[TeknoBug]

It sure can, not sure why you are stating it won't.

I have a 6570 with a failed fan and the computer shuts off completely due to overheating.

I've actually never had a PC shut off on an overheating GPU, my 560Ti's fans died and it skyrocketed to 112C and the PC became unresponsive but didn't shut off, not to mention an excessive amount of heat coming from the case. Oh you remember Nvidia 196.* drivers? I sure do, they killed GPU's with no warning because the driver set fan speed to 0%.

 

[tweakboy]

Great work interns LOL

 

[SolMiester]

I've actually never had a PC shut off on an overheating GPU, my 560Ti's fans died and it skyrocketed to 112C and the PC became unresponsive but didn't shut off, not to mention an excessive amount of heat coming from the case. Oh you remember Nvidia 196.* drivers? I sure do, they killed GPU's with no warning because the driver set fan speed to 0%.

You know, I have never heard anyone on these forums state their card died because of those drivers. They were on the site maybe a day, before being taken down...

 

[redsox0914]

Thanks for all the replies guys.

Even though I have two power supplies, one thing I realized with people pointing at the power supply is that I'm not changing is the cord connecting the power supply to the wall. Each setup has the same cord connecting it to the computer.

Is that another thing worth investigating (is it possible for one of those to "go bad"?)

 

[redsox0914]

So, OP, I suggest you troubleshoot this in a very disciplined manner by looking at all the facts you have in front of you:

Thanks for this list. Let me add or clarify things where appropriate.

(1) An Intel X79 system with an HD7970 and Rosewill 1000W power supply "died" a few weeks ago (you should specify whether the symptoms were identical to what you found later on a backup system, by the way).

Power button (both on the case and the motherboard) would not work. The motherboard lights (including the one on the CPU cooler) still lit up, but pressing power would not spin the fans or do anything.

(2) The HD7970 caused random system shutdowns on an AMD-based system using a generic power supply, whether in the BIOS, Windows, or a game.

This happened about two weeks later after putting it in the AMD system. Fans will spin (including on the GPU), computer will sound like it's on, but there is no signal to the monitors. Computer does not actually shut down with the 7970. It just never sends any display to the monitors.

(3) An open box R9 280X purchased for testing exhibited the exact same (confirm this) symptoms.

280X turns on, sometimes even going to Windows when I want it to, but it's the 280X that gets the random shutdowns.

(4) Using the Rosewill 1000W power supply in the AMD-based system did not alleviate the problems.

Confirmed

(5) An HD3870 video card works fine in the AMD-based system, with the generic power supply. Has not yet been tested in the Intel system.

Also works with the Rosewill 1000W psu.

Let me add one more condition that might make a difference.

The power cord from the PSU to the wall outlet is the same cord in both setups

There are three variables we're working with here: the video card, the power supply, and the motherboard

My original hypothesis was that between the motherboard and CPU on the Intel system, one of them had gone bad. But now people are starting to point at the CPU.

 

[Termie]

OK, it sounds like you might actually have two problems. Your Intel X79 would not boot. A system that will not boot has either a motherboard, RAM, or a PSU problem. It almost certainly isn't the CPU or the video card. CPU failures are incredibly rare, but you should let us know if there was any extreme overclocking on your 3930K, or perhaps a leak in your water cooler.

Second issue - the AMD system will boot, but will randomly shut down. That's a different problem all together. It apparently works with the HD3870, so it is not likely to be a motherboard or a CPU problem. It could be a video card problem, obviously, or it could be a PSU problem. Also note that the HD3870 uses a much older PCIe spec than the HD7970/280X. Let's not forget that in our troubleshooting. Your AMD motherboard appears to have been released two years before the HD7970. There could be an incompatibility there. Not a failed motherboard, but a compatibility issue. If you were really interested in running the HD7970 in that system, you'd want the latest BIOS, but your Intel system is the real focus here.

So back to that X79 system - I really would like you to try the HD3870 in that system. I know it means plugging the PSU back into it, but it's probably worth the trouble to narrow down the problem. If it does not boot with the HD3870, you have a motherboard issue, or perhaps a RAM issue - pull all but one stick out during the test to eliminate that variable as much as possible.

 

[alexruiz]

I would lean towards the video cards.

While rocketfish psus, most of them made by huntkey, are usually garbage, this is indeed a decent psu. Let me find the hardwaresecrets review of this psu, but the rocketfish should be good for crossfire hd7970s.

I say it is the video cards as those hd7970s / r9 280x are very fragile unbfortunately, I have lived it

As others suggested, use the hd3870 to diagnose both system, I am fairly sure all combinations of psu and mobo will work fine with the hd3870. You just got 2 bad video cards.

 

[redsox0914]

Well I got home real late last night so I didn't get to test anything out.

I took the video cards to a friend's place and threw them in his X58 system/ULTRA 750W PSU.

We flashed his BIOS to the latest version we could find and the results are basically identical to what I'm seeing in my AMD system.

7970 will not post or send a signal to the monitor.

280X will start but signal will be lost in about two minutes after power is on.

I'm pretty convinced now that both cards are dead.

I'm also pretty sure the Intel motherboard is also dead (how do I distinguish between dead motherboard and CPU again?) but the AMD system is still OK.

 

[Termie]

I wonder if you somehow had a short circuit that took out both the 7970 and the motherboard. Do you recall anything happening right before the failure, such as a noise, smoke, a liquid spill, extreme overclocking, etc.?

It is very difficult to damage a CPU. That's why you can assume it's the motherboard. But if you were doing extreme overclocking, then CPU damage is a possibility.

 

[redsox0914]

I don't think I ever took the CPU past 3.5, so overclocking shouldn't be a factor.

I definitely don't think any of the other stuff happened.

For what it's worth, the Intel board I have was originally RMA'd and ASUS had to send me a replacement board, so it doesn't surprise me that much that it'd be bad. EDIT: this was all a bit over 2 years ago.

But as far as taking out both, the 7970 was used (in the AMD board) for about two weeks after the Intel system died. I think I'm convinced now that the power supply is fine, so that would seem to make the board and the 7970 two separate events.


I've already submitted a refund request to Newegg for the 280X and have a (new) 770 ordered.

Sometime this weekend I'll dig up all the socket covers on the Intel board and try to get that RMA'd as well. Unfortunately I threw an aftermarket cooler on the 7970 so I think I'm SOL there.

 

[beginner99]

I don't think I ever took the CPU past 3.5, so overclocking shouldn't be a factor.

I definitely don't think any of the other stuff happened.

Maybe it was a power surge. How stable is power at your place? Have you ever had other non PC related stuff unexpectedly fail?

That the open box 280x has a defect doesn't seem too surprising. It was returned for some reason and then not properly tested. They guy in the shop might just have put it in a PC and booted and didn't wait those 2 minutes...So what Termie says makes sense. There probably was an event that caused both the MB and GPU to fail. Maybe one of them caused the other one to fail...

 

[redsox0914]

This was the first time I've seen something fail at this apartment (lived here since August).

As far as one causing the other, is it possible to have had something "hit" both but have the video card "live" about 2 more weeks before completely going?