Server execution runs 2_cuda_kernels_and_convolution.py on a fresh sandbox in your workspace. Sign in to launch one.