I've got a rig with 18 GPU's total (10 AMD 8GBs and 8 NVIDIA 6GBs) and I'm having a DAG problem. I have it set to run the AMD in one claymore instance, and the NVIDIA in another, and either can run on it's own just fine, but when I try to run both at once the machine turns off when it's trying to create the DAG files. With -gser 2 I can see that after all the AMD cards load, 5 of my 8 NVIDIA cards load, but when it gets to trying to do the 6th it dies (no specific card, it seems to load them randomly each time but it's always on whichever card is #6, which is part of why I think it's maxing something out). I've got a huge page file on a second SSD (tried 250GB on a separate SSD with no luck, then tried loading windows on a single 500GB SSD and using 250GB of it for the page file to see if I'd have better luck with it on a single drive) but it always fails around that 6th drive. I normally have 4GB of physical ram in my rigs, so I tried 8GB but it's still giving me this problem.
When it gets to the point to "create GPU buffer" is when it usually just completely dies and the machine just turns off, but sometimes it doesn't shut down and will give me a "calc DAG failed" error.
Would love any suggestions on how to prevent this if possible or if there's some limitation with claymore.
Hello!
Well ,first of all, switch from ETC to ETH ( if you're on ETC) to check DAG issues, i think for now DAG size of ETH is lower then ETC , my rigs with the seemslike troubles working on ETH for now.
Then , it was a lot of post then increasing phisical RAM solved alot of problems with DAG generetions. Sertainly 4 Gb is to low for those number of GPUs , suggest to try even 16 Gb.
And of couz try PM. CM is on its own way to dieing.
