Skip to content
GitLab
Commit 8ff58093 authored by Johann's avatar Johann Committed by Gerrit Code Review
Browse files

Merge "vp9: neon: convolve: replace some insns with simpler equivalents"

parents 8bebfbf7 355cb14d
v1.14.0-linphone 1.4.X experimental feature/update_to_v1.9.0-linphone feature/uwp_nuget forest frame_parallel highbitdepth indianrunnerduck javanwhistlingduck khakicampbell linphone linphone-android linphone-old longtailedduck m31-baseline m49-2623 m52-2743 m54-2840 m56-2924 m66-3359 m68-3440 mandarinduck mcw mcw2 nextgen nextgenv2 pcs-2013 playground sandbox/Jingning/experimental sandbox/Jingning/transcode sandbox/Jingning/vpx sandbox/aconverse@google.com/ansbench sandbox/debargha/playground sandbox/hkuang/frame_parallel sandbox/hkuang@google.com/decode sandbox/jimbankoski@google.com/proposed-aom sandbox/jingning@google.com/decoder_test_suite sandbox/jingning@google.com/experimental sandbox/jzern@google.com/test sandbox/wangch@google.com/vp9 sandbox/yaowu@google.com/mergeaom stable-vp9-decoder v1.12.0-linphone v1.6.1_linphone v1.7.0-linphone v1.9.0-linphone v1.9.0 v1.9.0-rc1 v1.8.2 v1.8.1 v1.8.0 v1.7.0 v1.6.1 v1.6.0 v1.5.0 v1.4.0 v1.3.0
No related merge requests found
Hide whitespace changes
Inline Side-by-side
Showing
with 10 additions and 16 deletions
vp9/common/arm/neon/vp9_convolve8_avg_neon.asm
+ 6
− 12
View file @ 8ff58093
......@@ -128,8 +128,8 @@ loop_horiz
vqrshrun.s32 d5, q15, #7
; saturate
vqshrn.u16 d2, q1, #0
vqshrn.u16 d3, q2, #0
vqmovn.u16 d2, q1
vqmovn.u16 d3, q2
; transpose
vtrn.16 d2, d3
......@@ -137,10 +137,7 @@ loop_horiz
vtrn.8 d2, d3
; average the new value and the dst value
vaddl.u8 q8, d2, d6
vaddl.u8 q9, d3, d7
vqrshrn.u16 d2, q8, #1
vqrshrn.u16 d3, q9, #1
vrhadd.u8 q1, q1, q3
vst1.u32 {d2[0]}, [r2], r3
vst1.u32 {d3[0]}, [r2], r3
......@@ -234,14 +231,11 @@ loop_vert
vqrshrun.s32 d5, q15, #7
; saturate
vqshrn.u16 d2, q1, #0
vqshrn.u16 d3, q2, #0
vqmovn.u16 d2, q1
vqmovn.u16 d3, q2
; average the new value and the dst value
vaddl.u8 q8, d2, d6
vaddl.u8 q9, d3, d7
vqrshrn.u16 d2, q8, #1
vqrshrn.u16 d3, q9, #1
vrhadd.u8 q1, q1, q3
vst1.u32 {d2[0]}, [r2], r3
vst1.u32 {d2[1]}, [r2], r3
......
vp9/common/arm/neon/vp9_convolve8_neon.asm
+ 4
− 4
View file @ 8ff58093
......@@ -120,8 +120,8 @@ loop_horiz
vqrshrun.s32 d5, q15, #7
; saturate
vqshrn.u16 d2, q1, #0
vqshrn.u16 d3, q2, #0
vqmovn.u16 d2, q1
vqmovn.u16 d3, q2
; transpose
vtrn.16 d2, d3
......@@ -213,8 +213,8 @@ loop_vert
vqrshrun.s32 d5, q15, #7
; saturate
vqshrn.u16 d2, q1, #0
vqshrn.u16 d3, q2, #0
vqmovn.u16 d2, q1
vqmovn.u16 d3, q2
vst1.u32 {d2[0]}, [r2], r3
vst1.u32 {d2[1]}, [r2], r3
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment