when I try to use cudaMalloc(),there is an errer 700 an illegal memory access was encountered with two graphics cards，but there is no error when I use one
TensorRT Version :126.96.36.199
GPU Type :Tesla T4
CUDNN Version :8.0.4
Operating System + Version :Ubuntu 18.04
It means that a kernel launch prior to the
cudaMalloc() call hit an error. Nothing else can be deduced from the provided information. You may wish to use a method such as this to get further clarity. For general CUDA debugging principles, you may wish to study session 12 here.