It appears to be a problem of too much overclocking and the cards do not support it. I have not any experience with Gtx 1080 ti but from my work experience up to 1070 the overclock they support if you are using Windows is +500 memory for the 1070 and plus 800 1060 and the 1050 ti goes way back to max stable overclock of +1000 memory.
All settings taken from Msi Afterburner. High temperature do not have anything to do with hashrate, I had a 1050 ti which was working for more than one day with 96 degree celsius because I didn't see that its fan stopped working completely and the hashrate was the same 15.4 mhs for it. I cleaned it completely and now is doing same hashrate with 63 degree celsius.
Topic: GPU Hashrate drops from 750 to ~300 suddenly (Read 222 times)
If you sure that they work few minutes at 750 Sol then GPUs are 100% fine. First of all try downclock them to see what will happen. And as other ppl say, check temps, maybe you need to renew thermal grease.
What about their temperature ? If not new GPUs, have you tried air blow gun on fans to remove dust ?
How are the fans controlled ?
How are the fans controlled ?
I doubt that OP's problem is temperature related, unless those temps rise to extreme levels, in that case maybe it is possible what you refer. If those 3 GPUs are not new, than there is possibility that they are damaged and that they can not give 100% of their memory load.
@OP, have you made some changes in their settings and tried to OC your GPUs, maybe you made a mistake in setting them up. Check software.
What about their temperature ? If not new GPUs, have you tried air blow gun on fans to remove dust ?
How are the fans controlled ?
How are the fans controlled ?
Ah.. this is a good suggestion. 
Yeah, power issue would be possible. OP try to follow this suggestion.
OP
That's why you need to describe your condition as details as possible. May you need to follow my old post about how to explain your troubled rig.
https://bitcointalksearch.org/topic/m.33080149
Yeah, power issue would be possible. OP try to follow this suggestion.
Well, if you’ve swapped cards around and those particular slots have the same issue, I’d suspect a power problem. Disconnect the power from a couple of working cards and see if the bad ones then work correctly.
OP
That's why you need to describe your condition as details as possible. May you need to follow my old post about how to explain your troubled rig.
https://bitcointalksearch.org/topic/m.33080149
Well, if you’ve swapped cards around and those particular slots have the same issue, I’d suspect a power problem. Disconnect the power from a couple of working cards and see if the bad ones then work correctly.
Have you fixed it?
If you have another motherboard, would be better if you tried those three cards in other board. It will be giving you more specific issue and would be easier to track what's the problems are.
Tell us your rig spec's (more detail), So, we can give you more analyze related to your problems.
If you have another motherboard, would be better if you tried those three cards in other board. It will be giving you more specific issue and would be easier to track what's the problems are.
Tell us your rig spec's (more detail), So, we can give you more analyze related to your problems.
Hi All,
I have 10 Zotac 1080 TI Extreme GPUs, running on windows 10, 64 bit.
Windows updates & Nvidia drivers are all the latest
I have 3 GPUs out of the 7, which start hashing at 750 Sol, but after few minutes they drop all the way down to ~300 or even ~250
I have tried different power risers, different PSUs, different PCIe slots, nothing has helped
The same 3 GPUs will always start at 750 and go down to ~300
The same 7 GPUs will start at 750 and stay at the same level, no matter how i swap the PSUs, powered risers, slots, sockets, etc...
The display is connected directly to the motherboard not to any GPU All GPUs has power state of P2, so i dont believe that one will do 750 and the other 250 because of this
I tried different miners, but the two showed absolutely the same result. 3 GPUs badly perform and 7 perform very well
I noticed that "memory controller load" on the high performing GPUs is about 80% while on the badly performing GPUs is around 55%, using Hardware Info software. Is this the root cause? Any idea how can i fix this?
I have 10 Zotac 1080 TI Extreme GPUs, running on windows 10, 64 bit.
Windows updates & Nvidia drivers are all the latest
I have 3 GPUs out of the 7, which start hashing at 750 Sol, but after few minutes they drop all the way down to ~300 or even ~250
I have tried different power risers, different PSUs, different PCIe slots, nothing has helped
The same 3 GPUs will always start at 750 and go down to ~300
The same 7 GPUs will start at 750 and stay at the same level, no matter how i swap the PSUs, powered risers, slots, sockets, etc...
The display is connected directly to the motherboard not to any GPU All GPUs has power state of P2, so i dont believe that one will do 750 and the other 250 because of this
I tried different miners, but the two showed absolutely the same result. 3 GPUs badly perform and 7 perform very well
I noticed that "memory controller load" on the high performing GPUs is about 80% while on the badly performing GPUs is around 55%, using Hardware Info software. Is this the root cause? Any idea how can i fix this?
Jump to: