What's the difference between 'ae_f32x2' and 'xtfloatx2' in SIMD optimization of 'xa_nn_matXvec_f32xf32_f32'?
Chen1399 opened this issue · 1 comments
Chen1399 commented
I found the 'xa_nn_matXvec_f32xf32_f32''s implementation uses 'xtfloatx2'. How about use 'ae_f32x2' to change it? Which one is faster?
nyadla-sys commented