Kaustubh Raste

@fbarchard @allenkelly @bhbruce @keeranroth, If you have time, please help review this PR.

@fbarchard @allenkelly @bhbruce @keeranroth can someone merge the changes?

@fbarchard @allenkelly @bhbruce, If you have time, please help review this PR.

@fbarchard @allenkelly @bhbruce @keeranroth, If you have time, please help review this PR.

@keeranroth I ran the mean-nd-test which uses f32 rsum on qemu and used ms calculated by xnnpack as reference for comparison. From comparison, i found the u4v is better than...

@keeranroth I updated the LMUL=2 code to have 8 accumulators in loop and taken performance again for mean-nd-test. From observations the LMUL=4 still takes less time (even when both have...

Hi, any more review comments on this patch?

Hi, any review comments? If not, can we get this committed please?

@fbarchard @allenkelly @bhbruce @keeranroth, If you have time, please help review this PR

If no review can we get this committed please?