Searched defs:numel_per_thread (Results 1 – 2 of 2) sorted by relevance
61 const size_t numel_per_thread = alignment / element_size; in init_elementwise_launch_config() local85 constexpr size_t numel_per_thread = alignment / sizeof(T); in multimem_all_reduce_kernel() local178 constexpr size_t numel_per_thread = alignment / sizeof(T); in multimem_one_shot_all_reduce_kernel() local
25 …uint64_t numel_per_thread = static_cast<uint64_t>(max_threads_per_block) * static_cast<uint64_t>(s… in getApplyGrid() local