Great idea and I am glad somebody actually tries to make an implementation.
But unfortunately this won't replace mining: the large scale mining farms have high end GPUs in their rigs, but the rest of the HW is very low end, because that works perfectly for mining and they want their ROI as low as possible.
I have a 6 GTX 1070 GPU rig, which would be decent GPUs for AI/ML, but the rest of the rig has a Celeron CPU, 4gb RAM,and a 5400 RPM HDD. Oh, and to be able to see all the GPUs, I had to downgrade all the PCIEx slots to 1x.
I am curious what kind of ML tasks would be able to fully utilize these GPUs, on such a low end hardware.
But unfortunately this won't replace mining: the large scale mining farms have high end GPUs in their rigs, but the rest of the HW is very low end, because that works perfectly for mining and they want their ROI as low as possible.
I have a 6 GTX 1070 GPU rig, which would be decent GPUs for AI/ML, but the rest of the rig has a Celeron CPU, 4gb RAM,and a 5400 RPM HDD. Oh, and to be able to see all the GPUs, I had to downgrade all the PCIEx slots to 1x.
I am curious what kind of ML tasks would be able to fully utilize these GPUs, on such a low end hardware.