Reading GPU memory from another process


I have two processes:

Process A C++ with cublas

Process B with python

The goal is to read data from process A in the GPU and train in process B

How can I start looking into that? My question is from the point of view of process B.

You could use CUDA IPC handles, but since Process A is a pure C++ code you would need to unpack the data manually as described here.

1 Like