onnxruntime
using GPU_WARP_SIZE to make kernel portable between AMD and Nvidia GPU
#5173
Merged
Merged