Skip to content

Commit

Permalink
Set transpose to false for unpack untilize
Browse files Browse the repository at this point in the history
  • Loading branch information
rtawfik01 committed Oct 23, 2024
1 parent dc92e25 commit b4f109c
Showing 1 changed file with 1 addition and 0 deletions.
1 change: 1 addition & 0 deletions llk_lib/llk_unpack_untilize.h
Original file line number Diff line number Diff line change
Expand Up @@ -77,6 +77,7 @@ inline void _llk_unpack_untilize_hw_configure_(const std::uint32_t unpack_src_fo
}

inline void _llk_unpack_untilize_init_(const std::uint32_t unpack_dst_format, const std::uint32_t tile_size, const std::uint32_t face_r_dim = FACE_R_DIM, const std::uint32_t num_faces = 4) {
cfg_reg_rmw_tensix<THCON_SEC0_REG2_Haloize_mode_RMW>(0);

const std::uint32_t unpA_ch1_x_stride = (unpack_dst_format&0x3) == (std::uint32_t) DataFormat::Float32 ? 4 : (unpack_dst_format&0x3) == (std::uint32_t) DataFormat::Float16 ? 2 : 1;
const std::uint32_t unpA_ch1_y_stride = FACE_R_DIM*unpA_ch1_x_stride;
Expand Down

0 comments on commit b4f109c

Please sign in to comment.