Mojo function

warp_sum

warp_sum[val_type: DType, simd_width: Int, //](val: SIMD[val_type, simd_width]) -> SIMD[val_type, simd_width]

warp_sum[intermediate_type: DType, *, reduction_method: ReductionMethod, output_type: DType](x: SIMD[type, size]) -> SIMD[output_type, 1]

Performs a warp level reduction using either a warp shuffle or tensor core operation. If the tensor core method is chosen, then the input value is cast to the intermediate type to make the value consumable by the tensor core op.

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