GPU Dead VRAM

[eggbox1]

I think I fried my VRAM on my rog strix 3070 while repasting it with k5 pro. Initially, there was artifacting, so I underclocked the GPU, which fixed the issue until I increased the memory clock. I checked the heatsink, everything  was making contact and I added more k5 pro for safety, but when I tried to push the memory clock again, the system crashed. Now there's no memory reported in gpu-z and severe artifacting. I'm unsure how this happened since everything seemed fine, possibly overheating of a VRAM chip. Any insights would be appreciated.iCloud Photos.zip

[ChrisLoudon]

1 hour ago, eggbox1 said:

Any insights would be appreciated

Hi buddy,

 

Yeh, artifacting is usually a sign of VRAM issues.

 

Personally, I prefer using thermal pads rather than paste but there shouldn't be any real issues other than the mess.

 

As it stands, I think the only thing you can do is clean all of the paste and give the chips a good inspection to see if there is any visible damage.

 

Get all the goop away then go over the rest with a soft toothbrush and distilled water.

 

Reseat the cooler with pads on the vram chips and vrm's and paste on the core die and see if the problem persists.

 

No much more I can recommend but you should probably prepare for the worst.

 

Good luck.

 

 

[eggbox1]

7 minutes ago, ChrisLoudon said:

Hi buddy,

 

Yeh, artifacting is usually a sign of VRAM issues.

 

Personally, I prefer using thermal pads rather than paste but there shouldn't be any real issues other than the mess.

 

As it stands, I think the only thing you can do is clean all of the paste and give the chips a good inspection to see if there is any visible damage.

 

Get all the goop away then go over the rest with a soft toothbrush and distilled water.

 

Reseat the cooler with pads on the vram chips and vrm's and paste on the core die and see if the problem persists.

 

No much more I can recommend but you should probably prepare for the worst.

 

Good luck.

 

 

I am 99% sure the gpu is dead, msi afterburner doesn't even detect it. I'll clear off the paste and have a look. So confused though because it must've died while it was under load.

[ChrisLoudon]

39 minutes ago, eggbox1 said:

So confused though because it must've died while it was under load.

Yeh, your probably right.

 

Strange for the VRAM to go fizz-bang as I was under the impression that 30 series and above VRAM modules had ECC of sorts so if you have pushed things too far, all you got was rubbish memory throughput.

 

Does the system post with the GPU installed and display an output to get you to a desktop or are you booting with another GPU / iGPU?

[eggbox1]

27 minutes ago, ChrisLoudon said:

Yeh, your probably right.

 

Strange for the VRAM to go fizz-bang as I was under the impression that 30 series and above VRAM modules had ECC of sorts so if you have pushed things too far, all you got was rubbish memory throughput.

 

Does the system post with the GPU installed and display an output to get you to a desktop or are you booting with another GPU / iGPU?

it posts with an output but windows displays that it's having issues in device manager and when I try to launch something simple like furmark it says it can't run because I need something with at least open gl 2.0

[ChrisLoudon]

38 minutes ago, eggbox1 said:

it posts with an output but windows displays that it's having issues in device manager and when I try to launch something simple like furmark it says it can't run because I need something with at least open gl 2.0

Ooft! A big pile of misery I think.

 

Not to throw any hope your way but there's a teeny tiny chance that there's maybe an impurity in the paste that's causing a short somewhere.

 

Give it a proper clean and see if anything changes.

[eggbox1]

2 minutes ago, ChrisLoudon said:

Ooft! A big pile of misery I think.

 

Not to throw any hope your way but there's a teeny tiny chance that there's maybe an impurity in the paste that's causing a short somewhere.

 

Give it a proper clean and see if anything changes.

I don't actually have any more k5 pro, should I run it with no thermal pads for like a super short time or use the old thermal pads?

[ChrisLoudon]

2 hours ago, eggbox1 said:

I don't actually have any more k5 pro, should I run it with no thermal pads for like a super short time or use the old thermal pads?

I dunno dude.

 

Maybe someone else can chime in with some advice on that??

 

I've booted a PC into BIOS for a short period with an dry EK block just resting on the CPU but not tried booting a GPU without any cooling.

 

The issue of course with GPU's is they are almost always upside down or vertical making it difficult to botch together a temp cooling solution when doing a bench build.

 

I would see if you can get a syringe of some thermal paste on Amazon on Prime delivery and use that on the core and pads on everything else.

 

If you have a riser cable you might be able to just rest the cooler without screwing it down but tbh, even although the card is probably dead, if it is something simple like a short you don't want to make matters worse.

[eggbox1]

3 hours ago, ChrisLoudon said:

I dunno dude.

 

Maybe someone else can chime in with some advice on that??

 

I've booted a PC into BIOS for a short period with an dry EK block just resting on the CPU but not tried booting a GPU without any cooling.

 

The issue of course with GPU's is they are almost always upside down or vertical making it difficult to botch together a temp cooling solution when doing a bench build.

 

I would see if you can get a syringe of some thermal paste on Amazon on Prime delivery and use that on the core and pads on everything else.

 

If you have a riser cable you might be able to just rest the cooler without screwing it down but tbh, even although the card is probably dead, if it is something simple like a short you don't want to make matters worse.

I cleaned up most of the goop and it’s still dead so yeah it’s proper gone. I think the card was a bit messed up from the start though, it was used and crashed when at stock clocks. I’m probably just gonna wait until amd releases a 4070 competitor and get that because gpus are crazy expensive at the moment. Thanks for your help!

[ChrisLoudon]

14 hours ago, eggbox1 said:

Thanks for your help!

No worries buddy.

 

Yeh, if it was playing up at stock then likely had an issue b4.

 

C