NOT real Temperature limit on GPU 1 reached, aborting...

[martin.po21]

Hy

Something wierd happens on one of my 5970 after rising gpu frequency to 780+ MHz...

On msi afterburner temperature graph i noticed that this card complete randomly reports around -703628469 degrees, and that triggers oclHashcat to stop.

It is always very below zero temperature.

Then after less than half second card reports normal temperatures.

I do not want to disable my watchdog, so i decided to ask what else can I do about it?
Should not watchdog be watching only positive temps?
Can it be set so it ignores "unrealsitic" changes?

Or is disabling watchdog and writting my own temperature control script my only option?

I understand that this is my cards fault, i am just asking for suggestions

Nontheless card runs just fine, but false temp reporting is slowly getting on my nerve

Thanks

[epixoip]

Try re-flashing the BIOS. There's actually two BIOS on the 5970, one for the master GPU and one for the slave GPU. Make sure you flash the correct BIOS to the correct GPU.

Been a while since I've ran 5970s, but this post helped me when I re-flashed mine:

http://forums.bit-tech.net/showthread.php?t=185918

I used the "800/1100 1.15v VDDC 1.15v VDDCI" BIOS found on that page, but you don't have to use one of those. The brand doesn't matter, by the way; AFAIK there were no OEM-design 5970s, they're all reference design, so a BIOS from any vendor will work. Before using the BIOS found on that page, I was running a Sapphire BIOS on an original ATi card (one actually manufactured by ATi, not produced by an OEM.)

If re-flashing the BIOS doesn't help then you might have a faulty temperature sensor or some other hardware malfunction.

[martin.po21]

Thanks, will try it out, hope it solves the problem

[martin.po21]

It turned out it was really BIOS problem.
I flashed BIOS from another properly working card and problem disappeared.
@epixoip thanks for suggestion!

[epixoip]

Not a problem, happy to hear it wasn't a hardware issue