Can other hardware destroy a GPU? +other confusing issues

I recently decided to upgrade my PC. Did my research and ordered the parts (Everything but the PSU, RAM and a secondary hard drive were brand new). After trying it out for a bit, I ran into a complete system freeze during gameplay (Nothing too fancy, just Mirror's edge on high graphics). After restarting the system the same happened during a YouTube video, but this time the PC couldn't restart and had several strange colors and shapes during boot-up, including large squares and horizontal lines. It booted, but around the point where the windows loading screen should've been, it rebooted again in an endless loop. Starting the PC in safe mode would get me to the desktop, but the screen was filled with horizontal lines which would disappear as if being erased in MS paint if I moved anything such as the mouse cursor over them. After disappearing, they'd just reappear somewhere else on the screen. The PC worked perfectly with an older GPU (Which I'm using now), so I sent the new one in for repairs.

Now, it's on it's way back, and I just want to be sure that it won't just blow up again when I put it back in. Any help as to how to make sure it won't happen again or what the weird lines mean or how a GPU managed to prevent the PC from booting would be much appreciated. I think the company I bought it from might get a little curious if I send the same product back twice in a short period of time.

GPU: GTX 780 (Blew up. Currently using Radeon HD 4850 as a replacement)
CPU: AMD FX-8320
Mobo: Asus M5A97 LE R2.0
RAM: A single Kingston 8Gb HyperX Blu, 1600MHz DDR3
PSU: SilverStone STT-ST70F-ES (700W)

Thanks in advance,

Discussion is locked
Reply to: Can other hardware destroy a GPU? +other confusing issues
PLEASE NOTE: Do not post advertisements, offensive materials, profanity, or personal attacks. Please remember to be considerate of other members. If you are new to the CNET Forums, please read our CNET Forums FAQ. All submitted content is subject to our Terms of Use.
Reporting: Can other hardware destroy a GPU? +other confusing issues
This post has been flagged and will be reviewed by our staff. Thank you for helping us maintain CNET's great community.
Sorry, there was a problem flagging this post. Please try again now or at a later time.
If you believe this post is offensive or violates the CNET Forums' Usage policies, you can report it below (this will not automatically remove the post). Once reported, our moderators will be notified and the post will be reviewed.
- Collapse -
I wouldn't read much into this.
- Collapse -
Seems normal enough

Seems normal enough for a GPU failure. Part of the nice under the hood improvements of Vista is the depreciation of the crappy old GDI+ drawing library in favor of something based on DirectX allowing for full GPU acceleration of even basic GUI features. Part of what Safe Mode does is revert to a software only rendering model, which is why it would work at least somewhat while normal booting wouldn't. As soon as Windows started trying to use some of those 3D functions, it triggered a failure in the video card causing a reboot.

Based on what you describe, it sounds like your video card just had a couple bad capacitors. As Bob said, it happens. This is what warranties are for, so if you're unlucky enough to get one that fails early, you can get a replacement.

- Collapse -

Mmm, I've been to a few different forums and this seems to be the gist of it. A few people think that the PSU might also be at fault, especially considering it's one of the parts that weren't replaced, and is slightly older. E.g. "The card and CPU have a combined 34A power draw. In a typical system, you are looking at roughly 3 to 4 amps in supporting components (motherboard, hard drive, DVD, etc). If you have more than one hard drive, and an additional 1A per drive. Depending on the case, between 1 and 2 more amps in fans. Worst case scenario, figure on 41~42A in use. You say you did not replace the PSU? Depending on the age of the power supply, you have to figure in degradation. It is easy to see how the system could be unstable with only 46A on tap - and I would certainly replace that PSU."

The card's coming back from repairs today, so the plan at the moment is to pop it in and keep an eye on the voltages using HWmonitor.

- Collapse -
I don't buy it

I don't buy it on the PSU end. The person you quoted was just trying to sound like they knew what they were talking about when it's pretty easy to tell they don't.

A PSU doesn't just have one big pool of power that it distributes out, like a wall socket or something. The person you quoted doesn't have even a basic understanding of Ohm's Law. It's true in the generic sense that a HDD and other components will put a drain on the PSU, but the number of amps a PSU can put out is dependent upon the power rail. If you look at the label on the side of the PSU it'll tell you exactly how many amps can be pushed through each power rail. You'll see a little chart and it'll have things like +12V and +3.3V then +12V might be say 24A, meaning there's a maximum of 24 amps that can be sent over the +12V rail (which is where your video card gets the bulk of its power). I won't even get into why they seem to think every drive would need 1A. They seem to be confused about what amps and volts are. Long story short, the person you quoted is one of those people who's really annoying to people like me. They read something someone says, only partially understand it, but don't let that stop them from rushing off to sound like some kind of genius by repeating it as many times as possible to others who know even less about the subject. And so a lot of bad information gets put out there and it's just more work to counteract it. So, if you didn't get much of the above, the long and short of it is, don't listen to the person who was on about the PSU, they're the type who pretends to know more than they actually do.

The PSU is unlikely as a culprit because if you were running close to the edge of what the PSU could handle, odds are safe boot wouldn't have worked either. The card is still active and drawing power. Not as much power, but it's still displaying an image so it's drawing some. You wouldn't have gotten the lines all over the screen if it were the PSU, unless you want to try and make an argument for dirty power and short of an oscilloscope readout, no one would likely believe that when there are far more probable explanations.

Back when I did Apple warranty repairs, the early model Mac Pros had the GeForce 7300GT card in them standard. These had a known defect with the two capacitors right at the solder points for the output ports and caused almost exactly the same issues you described. If you ran the older 2D software rendered diagnostics, it would work, but you'd get a bunch of lines all over the screen. If you tried running Mac OS X proper, which has an OpenGL based 3D GUI, things wouldn't go well. It had nothing to do with the PSU, but rather the power regulation on the card itself.

- Collapse -
The internet dilemma

Sometimes I wonder why I bother asking anything about anything on the internet. As far as I'm concerned, you could be talking just as much **** as you claim he is. With no first-hand experience I really don't know who to trust.

Well, the card came back from repairs and seems to be fine, so apparently the majority of people were right in saying that that's almost certainly the problem. Getting some fairly confusing voltage warnings from some software, but I'm completely clueless when it comes to the finer points in PSUs. AI Suite II is constantly showing voltages above those of the rails (e.g. the 3.3V rail is showing 3,312V atm), HWMonitor on the other hand is showing 3,147, and none of these programs show anything resembling amps.

People are judging my PSU based on how much it weighs, calling ******** on eachother's statements at every turn and I have almost as many opinions as I have people responding.

The internet's a terrifying place.

- Collapse -
Feel free to ask an electricical engineer

Feel free to ask an electrical engineer or computer engineer. I'd put money on the fact that they'd say I was right. I may have taken a few minor liberties to avoid getting bogged down in mundane details that'd cure even the most chronic insomnia, but I'm very true to the overall spirit. All you really need to do is ask yourself what that label on the side of the PSU is telling you if not exactly what I said. It's telling you that the PSU's power is spread out over these different power rails, so it's not like you have a pool of 100A (for example) and then every device connected to the PSU subtracts some number of amps from that pool. It's not that simple, which is why when you get a video card these days you can't just look at the total number of watts a PSU provides, you have to make sure it can also deliver enough amps to power the card.

And you won't really see anything measuring amps in general because most people, even hardcore overclocking types, don't really understand what it is they're looking at when they look at a voltage reading. The whole reason those readings are there is simply to make people feel impressed that they can monitor some statistic about their computer they don't even understand. It's a staple of marketing, where they list every single possible good aspect of some product right on the box. Like, you might know Energy Star is some kind of power saving standard, but do you really know the difference between say Energy Star 4.0 and 5.0 ratings? No, but if everything else was equal, odds are you'd pick a motherboard/computer with the 5.0 rating, wouldn't you?

In any event, good to hear the refurb video card resolved the issue.

- Collapse -
The irl dilemma

Honestly, I've seen how some of the electrical engineers are trained around here, and I trust most of them about as much as I trust random people on the internet. This might just be a local problem, but the people teaching around here are a scourge, and most of the things people seem to have learned are either from other people (who are just as eligible to be misinformed as the internet folk), the internet (again, same issue) or the teachers who are either teaching from old material which doesn't apply as much to the new technology, or are just generally uninterested in what's going on.
tl;dr: I don't trust people.

Not entirely sure how we got from PSUs to this misanthropic rant on trust, uninformed people and modern education. Anyhoo I've been having a chat on another forum about the power issue. The people there do seem knowledgeable, but again, what do I know. Be that as it may, I've been persuaded to buy a new PSU. Once I've found something I consider suitable, I'll probably pop back here just in case you happen to still be around

- Collapse -
It's probably not needed

It's probably not needed, but a new PSU won't hurt anything exactly either. The one you linked will probably work just fine. You don't say exactly which 780 card you have, but 58A on the 12V rail should be more than sufficient and it being an 80+ PSU is also a little added bonus. In large part the 80+ program is a bit of a marketing gimmick, since it doesn't guarantee that the same levels of efficiency are shared at all power loads, but at the same time you'd pretty much have to go out of your way to design a circuit that was very efficient only at specific power loads. There's bound to be some efficiency bleed across the board.

The main thing I'd make sure of is whether or not your video card needs one or two AUX power connectors and if they're 6 or 8-pin. Looking at the specs a bit closer it says 2x6+2 PCIe connectors, so that should have you covered as long as you're not looking to do SLI, in which case make sure you get a PSU rated for SLI. You don't want to know what happens if you try and pull power levels intended for 8 pins over a 6-pin connector, it's not pretty.

You might also give some consideration to a modular power supply. Mostly so you can remove whatever cables you aren't going to be using, like that floppy power connector for example. Then you don't have this giant rats nest of unused cables bunched up inside the case impeding air flow. If you can find something pretty similar to the PSU you linked to that's modular and not too much more expensive, I'd go with the modular myself. If it's like $100 more, then screw that, but under $50 I'd probably give it very strong consideration.

You also get bonus points for going with NewEgg since they've been one of the few major companies to take a stand against the plague of patent trolls in this country. I only wish I had the money and room to throw them more business.

- Collapse -
Wish I could go with NewEgg, but I can't

The pins I do have covered, and no, I'm not going for SLI. I've actually been working with non-modular PSUs all my life, so that'd probably be a very welcome change.
Well, the bad news is that after extensive research, I couldn't find any web store that would deliver the GS700 to where I live, so I'm back on the hunt. What I found from a local PC store was a 750W Corsair from the CSM series. 80+ gold, 61A on the 12V rail, claims to have 92% energy efficiency. NewEgg seems to list it as modular, but the local PC store I'm buying from claims it's semi-modular. Relatively cheap too. I was advised against buying from the CX or builders series, apparently those use low quality parts. Not entirely sure how I'm supposed to find out what's made out of quality parts and what's not, without just having the brand knowledge to begin with.

Anyhoo, here's the CS750M:

- Collapse -

CNET Forums