Author

Topic: Anyone have a 5970 with one REALLY hot VRM phase? (Read 2108 times)

full member
Activity: 186
Merit: 100
Nice font...
hero member
Activity: 518
Merit: 500
same thing happened to mine and it died  Cry

Just sent it in for RMA.

Claim it is their fault and you get bonus points if you ask for a 7990 !

This is the reason why the prices are going up for GPUs.

Miners pushing cards to much then demanding compensation ...
sr. member
Activity: 438
Merit: 250
same thing happened to mine and it died  Cry
hero member
Activity: 575
Merit: 500
Contact isn't the issue. I've mounted a few too many 5970 HSF to screw it up three times in a row. Each time the pads had imprints from the VRM ICs also. I wish it was a contact issue.

Just an imprint in the pad doesn't mean the pressure is high enough, the issue might just be the cooler giving uneven pressure.
hero member
Activity: 630
Merit: 500
Did you ever think that the temp sensor for that VRM is broken?
hero member
Activity: 697
Merit: 500
Try and sell it on eBay as a gaming card Roll Eyes

On another note : this kind of abuse is why I stay away from used GPUs.

I'm not going to add to that issue on eBay. It won't even play TF2 though as the VRM phase overheats at the loading screen.
hero member
Activity: 518
Merit: 500
Try and sell it on eBay as a gaming card Roll Eyes

On another note : this kind of abuse is why I stay away from used GPUs.
hero member
Activity: 697
Merit: 500
This happens when the VRM has -no- contact with whatever is meant to cool it. I know this because I had the same thing too when my 5870 fan died and tried a heatsink/fan without VRM cooling, was impossible to put any load at any decent frequency without frying it instantly, could only run really low frequency at like 100c.

Double check that ALL your VRMS have cooling. Get thermal pads if you have to.

Contact isn't the issue. I've mounted a few too many 5970 HSF to screw it up three times in a row. Each time the pads had imprints from the VRM ICs also. I wish it was a contact issue.
legendary
Activity: 1344
Merit: 1004
This happens when the VRM has -no- contact with whatever is meant to cool it. I know this because I had the same thing too when my 5870 fan died and tried a heatsink/fan without VRM cooling, was impossible to put any load at any decent frequency without frying it instantly, could only run really low frequency at like 100c.

Double check that ALL your VRMS have cooling. Get thermal pads if you have to.
hero member
Activity: 697
Merit: 500
Has anyone had a 5970 where one of the vddc phases for a core hits thermal shutdown at less than stock clocks? I'm talking, one phase at 115-120C and all the other phases on the card in the 50s. I remounted the stock HSF twice with new pads each time and I then mounted one of my spare EK waterblocks to the card.



Bottom GPU-Z, VDDC #2 is sitting at 72 Celsius with the core clocked down to 300MHz @ 0.95v just to keep it cool. At 375MHz and 0.95v it hovers around 95 Celsius. This card is at the end of a 6 card serially linked GPU waterblock chain so the incoming water is a bit warm at ~33-34C but compare the top GPU-Z. All my other 5970s look like that, cores in the 40s, VRMs in the low to mid 50s depending on where in the waterloop it sits.

The previous owner sent me a screenshot of the card running 800MHz on the core with an uptime of around 20 days right before he shipped this and another card to me. I trust that the owner didn't do a bait and switch. Just curious if anyone has seen this and if it is a sign that the phase is about to pop or if they've recovered it with something like a bake in the oven.
Jump to: