quadro p400 problem - fan randomly sticks at 100%

[Nimblor]

I have a quadro p400 that is working just fine helping with plex encoding. However, every 48 hours, randomly, the fan on it will kick up to 100%. Then it sticks there and is not able to be adjusted by any of the tuners (afterburner or asus). When it does this, there are no special processes running. Task manager and tuners both say load is zero, but afterburner says temp is over 80 and asus tuner (forget what it is called) says the temp is 101 (!). It is not.

Generally encoding does not even cause the fan to noticeably increase in speed. Also, I use the onboard graphics for display, there is no display connected to the card.


I have tried running malwarebytes and a did a full defender scan. Didn't find anything. I tried the different tuners and setting different profiles, they allowed me to control the fan speed but not when it maxes out. It seems like the tuners crash when that happens. It also occurs when there is no tuner installed. I reinstalled drivers using the delete old versions options. I have tried the card in a different slot and also reseated the heatsink with fresh thermal paste. I am really at a loss. I am considering drastic action, like unplugging the fan from the card and pointing another fan at it from inside the case.

Does anyone have any ideas on this?


Thanks!!!

 

[don256us]

You've done what I would have done. Without actually knowing anything, I'd guess you have a defective chip.

 

[EarthDog]

You can try to flash the BIOS again... maybe there's a glitch in the matrix? Perhaps it's a hw issue with the sensor or something? Perhaps bake it if a BIOS flash doesn't work?

 

[Nimblor]

You can try to flash the BIOS again... maybe there's a glitch in the matrix? Perhaps it's a hw issue with the sensor or something? Perhaps bake it if a BIOS flash doesn't work?

A windows reset did not fix it... I kept hoping there was some kind of driver shenanigans going on or something, but nothing that a reset could cure.

I have no idea how to go about re-flashing the bios. I did check and it has the most recent display port update. What is 'bake it?'

Because the problem is so random and I use the card for transcoding, I have not been able to determine if it is still functioning normally when it goes to full speed fan mode or not. I should probably find something to log its stats and leave it running until there is an event to see if there is anything unusual or if it is just a fan operation quirk. Maybe I will do some handbrake tests, normal encoding vs super loud mode encoding to see if the times are similar. I ordered a mini-display port to HDMI adapter to see if it artifacts or otherwise displays damage of being horked if it is used for video.

It seems like a dumb investment to get like a 60 mm fan and hotglue it to the card over the current fan and hook it to a motherboard fan header. Do video cards work without a fan attached?

Its frustrating.

 

[Nimblor]

I ran some tests and the card appears to be horked. I got some baselines for a handbrake encode using the card to help the process. Processor solo vs igpu vs quadro p400. I did an encode of a 1080p film in 5.1 to 720p x265 stereo. Initially the igpu and quadro showed similar results. Then about 5 minutes in the gpu temp and fan on the quadro ramped up. I watched on afterburner as the temp went up... 60C, 70, 80, I tried shutting the encoding off at that point but it was too late, the temp just spiked 90, then everything disappeared from the afterburner screen and the fan on the quadro was maxed out. The fan speed did not go down. This is the same behavior that afterburner exhibits when the card freezes with the fan on full, although without the temp spike that I have ever observed.

I used my finger as a temp probe/analog hork-o-meter on the back of the quadro card, it was very hot. Possibly horked

I have very little experience with video cards and my hork-o-meter is uncalibrated.

Is this what happens when they are dead?

 

[EarthDog]

Hork is like..........vomiting. Borked is like......broken.

Cards die in any number of ways... but now you're saying you can believe it's overheating? Earlier you said it wasn't... just making sure.

Did you try to reflash the BIOS? Do the oven trick (bake it - there are threads here covering that)?

 

[Nimblor]

Hork is like..........vomiting. Borked is like......broken.

Cards die in any number of ways... but now you're saying you can believe it's overheating? Earlier you said it wasn't... just making sure.

Did you try to reflash the BIOS? Do the oven trick (bake it - there are threads here covering that)?

Yes. It overheated by all means I have to detect such an issue. It was not overheating before. E.g. when the fan would go to max, the card was still cool to the touch. When I did the encoding test, it was very hot. The temp maxed out before the fan did, and when the fan got to max it just stuck. I will try the bios reflash. I am skeptical of the baking idea because it seems to be doing a good job of baking itself.

 

[EarthDog]

It wasn't before but is now screams sensor/bios issue or something. That doesn't make sense... it was a flash flag before, but now it's real...??

I will try the bios reflash. I am skeptical of the baking idea because it seems to be doing a good job of baking itself.

hardy har, lol!!!

In all seriousness, you bake it at higher temps then it runs soooooooooo....

Look up "the oven trick" here...

 

[Nimblor]

Bios flash yields promising results! 1080p x264 DD7.1 to 720p x265 stereo 1:37 in about 20 mins. !!!
Temp at 40C


View attachment 215149

Test #2 to see if it is borked .. i will have to wait to see if it randomly goes to max fan. It's the waiting game now.

 

[Nimblor]

The fan was at full blast again when I came home from work. It was doing no encoding. Time to fire up the oven!

But there are no artifacts when using it. Just this random boost up to 100%fan and then it is locked. If I have it set as the display driver, It locks the screen black, forcing a reboot.