Greetings my fellow enthusiasts!
I need some trouble shooting help. Below is my scattered thoughts on what happened and my thoughts and actions so far.
BumbleBee went down during the PrimeGrid Race recently. He currently is alive, but not well. He is not accepting decent GPUs at the moment.
He was running dual R9-280X's. They were well spaced, but the PSU is located as the bottom of the case, and is of such a high output that the fan doesn't feel the need to run continuously, allowing that heat to get drawn into the bottom GPU.
After the first crash, I finally noticed the PSU fan/heat problem. The bottom card would reach 99C and crash the system. This is a well ventilated case, with 3 120mm fans at max speed pushing air into the system. Removing the side cover did not reduce the temps. Bee would run about 3-5minutes and crash. So I broke out an oscillating pedestal fan, running off the mains and the temps dropped to 66C. Then crashed. And crashed. And crashed.
Corrupted software? Ok. Uninstall, restart, install latest drivers, restart, put a load on the GPUs: crash. Restart, no load on GPUs, crash. Remove one GPU: crash. Swap GPU: crash.
Thinking both GPUs are toast, I tried a third R9-280X, and 1-3 minutes: crash. ???? Single rail PSU, so not likely that the power system is at fault. EDIT: Actually a dual rail PSU, but the PCIe gets the 2nd rail to itself, 70amps/840watts.
Replace AMD cards with very low end Nvidia Quadro: Runs forever. Hmmmmmmm. A Quadro 4000, one that requires 6pin power: Runs forever. Dual Quadro 4000: Runs forever.
Ordinarily I'd just do a bunch of part swapping but either my parts are busy, or have already been tested, and left me scratching my head still.
My thoughts at this point, in order of likelihood:
1.) Software corruption. I was initially thinking Windows is corrupted, but then why would the Nvidia cards run fine? Other than the driver uninstallation/reinstallation, what's a man to do?
2.) PSU issue? Perhaps the double Quadro (pulling only 75W + 75W) wasn't enough to expose the problem? Maybe the fan WAS supposed to be spinning, and I'm thinking of some other unit that has the temp controlled fan. It's Thermaltake brand, 1375 Watts. My past experience is that their products are (somewhat) innovative/stylish, but not of the best quality. Decent quality, but not the best.
3.) And the least likely suspect is that the 2 previously installed 280Xs are BOTH dead, not just the over-heated one, and that my working spare was dead before I installed it.
And with that long and drawn out story, I now conclude by asking for YOUR thoughts.
Thanks in advance, and Merry CHRISTmas!
Tony.
I need some trouble shooting help. Below is my scattered thoughts on what happened and my thoughts and actions so far.
BumbleBee went down during the PrimeGrid Race recently. He currently is alive, but not well. He is not accepting decent GPUs at the moment.
He was running dual R9-280X's. They were well spaced, but the PSU is located as the bottom of the case, and is of such a high output that the fan doesn't feel the need to run continuously, allowing that heat to get drawn into the bottom GPU.
After the first crash, I finally noticed the PSU fan/heat problem. The bottom card would reach 99C and crash the system. This is a well ventilated case, with 3 120mm fans at max speed pushing air into the system. Removing the side cover did not reduce the temps. Bee would run about 3-5minutes and crash. So I broke out an oscillating pedestal fan, running off the mains and the temps dropped to 66C. Then crashed. And crashed. And crashed.
Corrupted software? Ok. Uninstall, restart, install latest drivers, restart, put a load on the GPUs: crash. Restart, no load on GPUs, crash. Remove one GPU: crash. Swap GPU: crash.
Thinking both GPUs are toast, I tried a third R9-280X, and 1-3 minutes: crash. ???? Single rail PSU, so not likely that the power system is at fault. EDIT: Actually a dual rail PSU, but the PCIe gets the 2nd rail to itself, 70amps/840watts.
Replace AMD cards with very low end Nvidia Quadro: Runs forever. Hmmmmmmm. A Quadro 4000, one that requires 6pin power: Runs forever. Dual Quadro 4000: Runs forever.
Ordinarily I'd just do a bunch of part swapping but either my parts are busy, or have already been tested, and left me scratching my head still.
My thoughts at this point, in order of likelihood:
1.) Software corruption. I was initially thinking Windows is corrupted, but then why would the Nvidia cards run fine? Other than the driver uninstallation/reinstallation, what's a man to do?
2.) PSU issue? Perhaps the double Quadro (pulling only 75W + 75W) wasn't enough to expose the problem? Maybe the fan WAS supposed to be spinning, and I'm thinking of some other unit that has the temp controlled fan. It's Thermaltake brand, 1375 Watts. My past experience is that their products are (somewhat) innovative/stylish, but not of the best quality. Decent quality, but not the best.
3.) And the least likely suspect is that the 2 previously installed 280Xs are BOTH dead, not just the over-heated one, and that my working spare was dead before I installed it.
And with that long and drawn out story, I now conclude by asking for YOUR thoughts.
Thanks in advance, and Merry CHRISTmas!
Tony.
Last edited:
Facebook
Twitter