[x264-devel] [Git][videolan/x264][master] aarch64: Fix the zigzag_interleave_8x8_cavlc_neon function

[Martin Storsjö]

Martin Storsjö pushed to branch master at VideoLAN / x264


Commits:
55d517bc by Martin Storsjö at 2021-04-12T09:54:56+03:00
aarch64: Fix the zigzag_interleave_8x8_cavlc_neon function

Use 'cmhs' (which does an unsigned greater or equal comparison)
instead of 'cmhi' (which does an unsigned greater comparison).

This makes sure that dct coeffs with a magnitude of 1 are recognized
in the output nnz buffer.

- - - - -


1 changed file:

- common/aarch64/dct-a.S


Changes:

=====================================
common/aarch64/dct-a.S
=====================================
@@ -707,7 +707,7 @@ function zigzag_interleave_8x8_cavlc_neon, export=1
     umaxp      v16.8h, v16.8h, v18.8h
     st1        {v2.8h}, [x0],  #16
     st1        {v6.8h}, [x0],  #16
-    cmhi       v16.4s, v16.4s, v31.4s
+    cmhs       v16.4s, v16.4s, v31.4s
     st1        {v3.8h}, [x0],  #16
     and        v16.16b, v16.16b, v31.16b
     st1        {v7.8h}, [x0],  #16



View it on GitLab: https://code.videolan.org/videolan/x264/-/commit/55d517bc4569272a2c9a367a4106c234aba2ffbc

-- 
View it on GitLab: https://code.videolan.org/videolan/x264/-/commit/55d517bc4569272a2c9a367a4106c234aba2ffbc
You're receiving this email because of your account on code.videolan.org.