onnxruntime

abdbb5fc - Reduction kernel optimization (#6088)

Reduction kernel optimization (#6088) Optimize reduction kernel code by moving loads from global memory before computation. Add CMake option to build CUDA code with --generate-line-info option.