Re:GPU Open Analytics Initiative
by sgt101 on Saturday May 13, 2017 @10:19AM
(#14331094)
Hello - Interesting to think this through! The GPU has its own bus and I think sends data to its cores according to the CUDA instructions that it is given.
I'm hazy about how instructions get to the GPU... I guess the cpu us is executing the OS and the then gets instructions to load the program from disk and then into the execution space in the GPU's memory. I think that the CPU then pushes data into the GPU's memory space for execution and reads results out from the GPU memory according to synchronisation instructions in its program?
I'm hazy about how instructions get to the GPU... I guess the cpu us is executing the OS and the then gets instructions to load the program from disk and then into the execution space in the GPU's memory. I think that the CPU then pushes data into the GPU's memory space for execution and reads results out from the GPU memory according to synchronisation instructions in its program?