Vega 56 HBM Temp reading go crazy.

[Type 2501]

I just got myself a second handed Vega 56 AIB card. While I am having fun tweaking the voltage and clock. The HBM can't go anywhere above 915mHz without artefact, and the coil whine is very bad (I think it is about the HBM, because every time I change the HBM volt and hit apply, the whine will stop for half a second) at first I thought I just lose the lottery on HBM.

 

But then I notice something is very wrong with the HBM temperature reading from GPU-Z: 

The HBM temp reading is jumping between normal (50 to 70) and over 1000 degree! The image below is taken when the GPU running at balanced profile with Hevan benchmark running on background. 

 

Nomal:

 

Max HBM Temp:

 

I google a bit and found no other case like that, so there is very little chance it is the problem of AMD drivers or GPU-Z. Does anyone of you seen something like that?

 

I really should RMA this card, lucky that it still under warranty. 

[normpearii]

Looks like a bad temp sensor.


If you can still RMA it, then I would suggest just doing so.

[GamerDude]

Bah, I sometimes get crazy GPU and/or memory temp readings for my Vega 64's, I simply ignore such readings. 

[MrTiC]

Its a bug, i recently had my vega 64 running at 2300 mhz

[Zando_]

26 minutes ago, GamerDude said:

Bah, I sometimes get crazy GPU and/or memory temp readings for my Vega 64's, I simply ignore such readings. 

Yep, it's reported over 800,000C temps on the Vega FE I had, needless to say I didn't believe it since my house wasn't a smoldering crater. 

[Type 2501]

8 hours ago, Zando Bob said:

...

800,000C...

 

Lol...

 

9 hours ago, MrTiC said:

Its a bug, i recently had my vega 64 running at 2300 mhz

It is a bug from the lunch of vega, didn't they fixed that already?

 

 

BTW, Wattman is really broken as a software, sometimes it didn't do anything, and sometimes it just does whatever it wants. not to mention it sometimes crushed while working on it.

Since afterburner lack some features, is there any other tools we can use?

[Wormhole]

AMD is buggy my cpu showed 256c water cooled nothing to worry about

[Type 2501]

And today while I am playing ace combat 7 after work, the power draw spike to 1022w.

 

I am really thinking if I can use that as an excuse to rma mine and try my lottery luck once again. Since the one on my hand is a loser on OC.