Fix mismatch issue with InitAccOpt + InnerUnroll

ROCm / Tensile

Stretching GPU performance for GEMMs and tensor contractions.

MIT License

218 stars 147 forks source link

Fix mismatch issue with InitAccOpt + InnerUnroll #1858

Closed nakajee closed 9 months ago

nakajee commented 9 months ago

use const 0 for src1 of MFMFA only if index of innerUnrll (iui) is 0

nakajee commented 9 months ago

Some CI test fail, but there are known issues. 1) host library test fail in precheckin 2) fails with source kernels in extended (1sum_simple, 1sum_gsu_simple)