Mojo function

group_norm_gpu_multi_block_stats

group_norm_gpu_multi_block_stats[StatsLayoutType: TensorLayout, stats_origin: MutOrigin, //, dtype: DType, simd_width: UInt, input_fn: fn[width: Int](row: Int, col: Int) capturing -> SIMD[dtype, width]](stats: TileTensor[get_accum_type[dtype](), StatsLayoutType, stats_origin], num_splits: Int, group_size: Int)

Multi-block stats kernel: computes partial Welford statistics per split.

Grid: num_rows * num_splits blocks. Each block handles one split of one group and writes partial (mean, m2, count) to the stats buffer. Stats layout: stats[block_idx * 3 + {0,1,2}] = {mean, m2, count}.

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit