minimize gpuAtomicAdd overhead in bounds_check_indices_kernel_v2#5124

Open

liligwu wants to merge 5 commits intopytorch:mainfrom

ROCm:lii/opt_bounds_check_v2

Commits on Nov 12, 2025

opt embedding_bounds_check_v2
liligwu
committed

Commits on Nov 13, 2025

Commits on Nov 21, 2025

guard atomic operation optimization in /workspace/FBGEMM/fbgemm_gpu/codegen/utils/embedding_bounds_check_v2.cu by #ifdef USE_ROCM
liligwu
committed