[SOLVED] Checking if Rx 580 is dead?

[Greenc0102]

Hello all, just looking to make one last check on my Rx 580 before I accept that it has seen it's last day.

Yesterday I turned on my computer excitedly to download halo infinite, closed the door and went to work. This is one of the the only times I've ever left it running while I was not home. I got home last night to an unresponsive box. Pressing the power button would yield a click, a half turn of the fans and then a full cut out that needed the power to be turned off and on to undo. So I pulled it apart and unplugged everything from the PSU while preying that would be the culprit. Unfortunately I found the PSU to be innocent as it sprung to life with nothing else connected. One by one I reconnected power cables, testing as I went. Sure enough as my lucky last component (GPU) was plugged in, once again I was met with the dreaded, click, spin, off.

So luckily I live in a house of fellow nerds and I optimistically asked one if I could install my GPU in their rig to see if there was something dodgy going on with my motherboard or PCIE slot. However once connected up their computer did the same thing.

Now in one final effort to not buy a GPU in today's less than ideal market, I'm asking you guys if there's anything I can do, any checks I can run, to make sure I'm not replacing a salvageable gpu needlessly. I can't see any sign of damage on the GPU (bulging capacitors etc) or smell any of the usual smells that come with technology deciding it no longer wants to function.

Is it possible something happened to the power connector on the GPU? Also is there any way I can make sure my PSU is not the culprit and won't fry the next GPU I put in.

Thanks in advance for any help guys

 

[JohnBonhamsGhost]

a dead GPU shouldn't prevent a system from powering on.
the only symptom should be a blank screen or "No Signal" message.

is there any way I can make sure my PSU is not the culprit and won't fry the next GPU I put in.

provide your complete system specs along with how long the power supply has been in use.

 

[Greenc0102]

Motherboard: b450m mortar max
CPU: Ryzen 7 3800X
PSU: Gigabyte g750h
Ram: 32gb ddr4 gskill (unsure exact model I'm out of the house so can't check)
Graphics: MSI Radeon RX 580 ARMOR 8G OC

Power supply is probably 2 years old give or take. I purchased the system second hand last august.

I was hoping it was the power supply but the exact same thing happened when I put the gpu in my mates computer.

 

[Math Geek]

the best test is what you already did. which is testing it in another system. same symptoms in a different system obviously points the finger at the gpu.

i have never seen this happen where it did not mean the gpu was dead. there might be some fixable issue that has happened at some point in the past to cause such an issue, but i have never actually seen it

sadly looks like you are now in the market for a new gpu.

just to be complete about it, you can try another gpu in your system. hopefully a friend has one you can test with or you have an old spare one lying about. this will make you 100% sure it is your gpu when your system works with another gpu installed.

 

[Greenc0102]

the best test is what you already did. which is testing it in another system. same symptoms in a different system obviously points the finger at the gpu.

i have never seen this happen where it did not mean the gpu was dead. there might be some fixable issue that has happened at some point in the past to cause such an issue, but i have never actually seen it

sadly looks like you are now in the market for a new gpu.

just to be complete about it, you can try another gpu in your system. hopefully a friend has one you can test with or you have an old spare one lying about. this will make you 100% sure it is your gpu when your system works with another gpu installed.

 

[Greenc0102]

It's looking that way

Yeah I've been trying another GPU. Everything boots up except it's a relic from the past (2011 GTX 580) and only takes VGA output. I threw out my last VGA cable earlier this year and now feel like a fool.

Well I guess I'll mortgage my car or something and upgrade. Thanks for the help guys. As a final question, is there anything I can do with an old GPU? People who may be able to refurb it that I could sell it on to? Or at the very least where in Australia can you dispose of these things properly?

 

[JohnBonhamsGhost]

People who may be able to refurb it that I could sell it on to?

you'll find items like this on eBay and others sold as "for parts only".
maybe take a quick look around and see if you can find a good price range to start with.

 

[InvalidError]

Since you have an obvious short to ground on the GPU, it wouldn't be too difficult to hunt down the most likely dead parts (capacitors and FETs), knock the likely dead ones off the board and see if the short goes away. If all the Vcore/Vmem coils have been removed and the Vcore/Vmem short to ground persists, then the GPU die itself is fried.

With some "luck", the culprit will be obvious in visual inspection and save a lot of guesswork. (Luck in quotations here as visual cues typically come in the form of charred PCB which is much harder to repair since all of the charred stuff needs to be gouged out first to prevent shorts between layers.)

 

[Math Geek]

i wish that was something i knew how to do

so many times i've replaced gpu's wondering if a cap/fet could be replaced and save the whole thing. ....

but alas this is outside my knowledge base

 

[Greenc0102]

Since you have an obvious short to ground on the GPU, it wouldn't be too difficult to hunt down the most likely dead parts (capacitors and FETs), knock the likely dead ones off the board and see if the short goes away. If all the Vcore/Vmem coils have been removed and the Vcore/Vmem short to ground persists, then the GPU die itself is fried.

With some "luck", the culprit will be obvious in visual inspection and save a lot of guesswork. (Luck in quotations here as visual cues typically come in the form of charred PCB which is much harder to repair since all of the charred stuff needs to be gouged out first to prevent shorts between layers.)

Is there any tests I can do with a multimeter? Nothing obvious is out of place, no burns or anything of the like. I've bitten the bullet and bought an upgrade anyway but still would love to have a punt at salvaging it.

 

[InvalidError]

Is there any tests I can do with a multimeter? Nothing obvious is out of place, no burns or anything of the like. I've bitten the bullet and bought an upgrade anyway but still would love to have a punt at salvaging it.

For hunting down a non-obvious dead short, you start by testing resistance and diode voltage drop from all GPU supply rails to ground to find the one that appears to be shorted (you need a 0.01 ohm resolution for this since Vcore can be sub-ohm) and once you have found the problematic rail, you can either use a micro-ohm meter to find the component(s) around which the resistance value is lowest or attach wires to feed external power directly into the rail and see what gets hot. (need a power supply that can reliably deliver 2-5A at 0.5-1V without overshoots as those might kill components.) You could also feed 1A into the shorted rail and then use your meter on its lowest-voltage DC range and find the component with the lowest voltage drop, 1mV = 1mOhm.