Crashing under load - two R9 295x2 GPUs, one motherboard

[bbOOmm]

I have two XFX R9 295x2 GPUs plugged into the 16x slots on an MSI 970A-G46 motherboard.

The system will boot up and operate fine. Mining ethereum - I can load the DAG and the second that it starts mining with the stock clocks on the GPUs, One of the cards will litterally power down ( lights out)  while the other crashes and gives me a black screen. Upon further testing, I can downclock and all 4 GPU cores will run, BUT, I can only run the GPU's at around 650Mhz GPU clock and 650Mhz Mem clock for all 4 cores. If I try to increase any further, once again, I'll crash.

If I just use one R9 295x2 card, I can go stock clocks, plus over clocking.

The motherboard is on its own 650W power supply while the GPUs have their own 2980W power supply, so the GPUs have plenty of 12V, I'm just wondering if the PCI-e interface is the limiting factor that is causing the cards to crash... IE: not enough current on the buss

This motherboard is a gaming motherboard, however I have never been much of an overclocker since I really don't play games, so going into the BIOS to tweak settings is abit new. Can I change anything in the BIOS to make this run the two R9 295x2 GPUs???

[1715194837]

1715194837

[1715194837]

1715194837

[1715194837]

1715194837

[1715194837]

1715194837

[1715194837]

1715194837

[1715194837]

1715194837

[antantti]

I am currently running two 295x2 on MSI 970A-G43 mobo with zero problems, powered by 2 x rm1000.

Have you tried with only one card? Mobo bios updated?

[RentGPU]

I have two XFX R9 295x2 GPUs plugged into the 16x slots on an MSI 970A-G46 motherboard.

The system will boot up and operate fine. Mining ethereum - I can load the DAG and the second that it starts mining with the stock clocks on the GPUs, One of the cards will litterally power down ( lights out)  while the other crashes and gives me a black screen. Upon further testing, I can downclock and all 4 GPU cores will run, BUT, I can only run the GPU's at around 650Mhz GPU clock and 650Mhz Mem clock for all 4 cores. If I try to increase any further, once again, I'll crash.

If I just use one R9 295x2 card, I can go stock clocks, plus over clocking.

The motherboard is on its own 650W power supply while the GPUs have their own 2980W power supply, so the GPUs have plenty of 12V, I'm just wondering if the PCI-e interface is the limiting factor that is causing the cards to crash... IE: not enough current on the buss

This motherboard is a gaming motherboard, however I have never been much of an overclocker since I really don't play games, so going into the BIOS to tweak settings is abit new. Can I change anything in the BIOS to make this run the two R9 295x2 GPUs???

You are operating through two diffrenet power supplies seems to be something wrong with the wiring can you tell me how you managed the wiring between both

[bbOOmm]

I am currently running two 295x2 on MSI 970A-G43 mobo with zero problems, powered by 2 x rm1000.

Have you tried with only one card? Mobo bios updated?

Ok, I updated the bios to the latest, but still I cannot get both gpus to run correctly. I was able to tweak the clocks to 825/825 which gives me around 90MH, basically 3 GPUs worth of hashrate... out of curiousity, I added -t 3 to the ethminer command to run 3 of the 4 GPUs, and sure enough, it ran at stock clocks , but the minute I tweaked the clocks, poof, it crashes.

What is more troublesome, I added a pair of powered PCI-e extenders so the cards were not drawing any power off the PCI-e slots on the motherboard, that didn't work at all. That setup worked just fine with this one board I had, some cheap ass duo core, but alas, I accidentially cooked that board.... now I cannot get any other board to run both GPUs either directly plugged into the motherboard or through extenders.

So, I'm thinking the motherboard's PCI-e slot power is getting weak