Skip to content

L
libyuv
Commit 0a3d23c8 authored by Frank Barchard's avatar Frank Barchard
Browse files
Options

fix clang-format-ing for row arm functions 

TBR=kjellander@chromium.org
BUG=None
TEST=git cl lint

Change-Id: I45ecd7f8279981ba037dc051f521f6b6d5506f64
Reviewed-on: https://chromium-review.googlesource.com/664345
Commit-Queue: Frank Barchard <fbarchard@google.com>
Reviewed-by: 's avatarFrank Barchard <fbarchard@google.com>
parent 753a91cb
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 13 additions and 19 deletions
source/compare_neon.cc
View file @ 0a3d23c8
...@@ -29,7 +29,7 @@ uint32 HammingDistance_NEON(const uint8* src_a, const uint8* src_b, int count) { ...@@ -29,7 +29,7 @@ uint32 HammingDistance_NEON(const uint8* src_a, const uint8* src_b, int count) {
asm volatile( asm volatile(
"vmov.u16 q4, #0 \n" // accumulator "vmov.u16 q4, #0 \n" // accumulator
"1: \n" "1: \n"
"vld1.8 {q0, q1}, [%0]! \n" "vld1.8 {q0, q1}, [%0]! \n"
"vld1.8 {q2, q3}, [%1]! \n" "vld1.8 {q2, q3}, [%1]! \n"
"veor.32 q0, q0, q2 \n" "veor.32 q0, q0, q2 \n"
...@@ -60,7 +60,7 @@ uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count) { ...@@ -60,7 +60,7 @@ uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count) {
"vmov.u8 q9, #0 \n" "vmov.u8 q9, #0 \n"
"vmov.u8 q11, #0 \n" "vmov.u8 q11, #0 \n"
"1: \n" "1: \n"
"vld1.8 {q0}, [%0]! \n" "vld1.8 {q0}, [%0]! \n"
"vld1.8 {q1}, [%1]! \n" "vld1.8 {q1}, [%1]! \n"
"subs %2, %2, #16 \n" "subs %2, %2, #16 \n"
......
source/compare_neon64.cc
View file @ 0a3d23c8
...@@ -27,7 +27,7 @@ uint32 HammingDistance_NEON(const uint8* src_a, const uint8* src_b, int count) { ...@@ -27,7 +27,7 @@ uint32 HammingDistance_NEON(const uint8* src_a, const uint8* src_b, int count) {
asm volatile( asm volatile(
"movi v4.8h, #0 \n" "movi v4.8h, #0 \n"
"1: \n" "1: \n"
"ld1 {v0.16b, v1.16b}, [%0], #32 \n" "ld1 {v0.16b, v1.16b}, [%0], #32 \n"
"ld1 {v2.16b, v3.16b}, [%1], #32 \n" "ld1 {v2.16b, v3.16b}, [%1], #32 \n"
"eor v0.16b, v0.16b, v2.16b \n" "eor v0.16b, v0.16b, v2.16b \n"
...@@ -55,7 +55,7 @@ uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count) { ...@@ -55,7 +55,7 @@ uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count) {
"eor v17.16b, v17.16b, v17.16b \n" "eor v17.16b, v17.16b, v17.16b \n"
"eor v19.16b, v19.16b, v19.16b \n" "eor v19.16b, v19.16b, v19.16b \n"
"1: \n" "1: \n"
"ld1 {v0.16b}, [%0], #16 \n" "ld1 {v0.16b}, [%0], #16 \n"
"ld1 {v1.16b}, [%1], #16 \n" "ld1 {v1.16b}, [%1], #16 \n"
"subs %w2, %w2, #16 \n" "subs %w2, %w2, #16 \n"
......
source/row_neon.cc
View file @ 0a3d23c8
This diff is collapsed.
source/row_neon64.cc
View file @ 0a3d23c8
This diff is collapsed.
source/scale_neon64.cc
View file @ 0a3d23c8
...@@ -77,11 +77,9 @@ void ScaleRowDown2Box_NEON(const uint8* src_ptr, ...@@ -77,11 +77,9 @@ void ScaleRowDown2Box_NEON(const uint8* src_ptr,
"subs %w3, %w3, #16 \n" // 16 processed per loop "subs %w3, %w3, #16 \n" // 16 processed per loop
"uaddlp v0.8h, v0.16b \n" // row 1 add adjacent "uaddlp v0.8h, v0.16b \n" // row 1 add adjacent
"uaddlp v1.8h, v1.16b \n" "uaddlp v1.8h, v1.16b \n"
"uadalp v0.8h, v2.16b \n" // row 2 add adjacent + "uadalp v0.8h, v2.16b \n" // += row 2 add adjacent
// row1
"uadalp v1.8h, v3.16b \n" "uadalp v1.8h, v3.16b \n"
"rshrn v0.8b, v0.8h, #2 \n" // downshift, round and "rshrn v0.8b, v0.8h, #2 \n" // round and pack
// pack
"rshrn2 v0.16b, v1.8h, #2 \n" "rshrn2 v0.16b, v1.8h, #2 \n"
"st1 {v0.16b}, [%2], #16 \n" "st1 {v0.16b}, [%2], #16 \n"
"b.gt 1b \n" "b.gt 1b \n"
...@@ -101,7 +99,7 @@ void ScaleRowDown4_NEON(const uint8* src_ptr, ...@@ -101,7 +99,7 @@ void ScaleRowDown4_NEON(const uint8* src_ptr,
(void)src_stride; (void)src_stride;
asm volatile( asm volatile(
"1: \n" "1: \n"
"ld4 {v0.8b,v1.8b,v2.8b,v3.8b}, [%0], #32 \n" // src line 0 "ld4 {v0.8b,v1.8b,v2.8b,v3.8b}, [%0], #32 \n" // src line 0
"subs %w2, %w2, #8 \n" // 8 processed per loop "subs %w2, %w2, #8 \n" // 8 processed per loop
"st1 {v2.8b}, [%1], #8 \n" "st1 {v2.8b}, [%1], #8 \n"
"b.gt 1b \n" "b.gt 1b \n"
...@@ -230,7 +228,7 @@ void ScaleRowDown34_1_Box_NEON(const uint8* src_ptr, ...@@ -230,7 +228,7 @@ void ScaleRowDown34_1_Box_NEON(const uint8* src_ptr,
asm volatile( asm volatile(
"movi v20.8b, #3 \n" "movi v20.8b, #3 \n"
"add %3, %3, %0 \n" "add %3, %3, %0 \n"
"1: \n" "1: \n"
"ld4 {v0.8b,v1.8b,v2.8b,v3.8b}, [%0], #32 \n" // src line 0 "ld4 {v0.8b,v1.8b,v2.8b,v3.8b}, [%0], #32 \n" // src line 0
"ld4 {v4.8b,v5.8b,v6.8b,v7.8b}, [%3], #32 \n" // src line 1 "ld4 {v4.8b,v5.8b,v6.8b,v7.8b}, [%3], #32 \n" // src line 1
"subs %w2, %w2, #24 \n" "subs %w2, %w2, #24 \n"
...@@ -279,7 +277,7 @@ void ScaleRowDown38_NEON(const uint8* src_ptr, ...@@ -279,7 +277,7 @@ void ScaleRowDown38_NEON(const uint8* src_ptr,
(void)src_stride; (void)src_stride;
asm volatile( asm volatile(
"ld1 {v3.16b}, [%3] \n" "ld1 {v3.16b}, [%3] \n"
"1: \n" "1: \n"
"ld1 {v0.16b,v1.16b}, [%0], #32 \n" "ld1 {v0.16b,v1.16b}, [%0], #32 \n"
"subs %w2, %w2, #12 \n" "subs %w2, %w2, #12 \n"
"tbl v2.16b, {v0.16b,v1.16b}, v3.16b \n" "tbl v2.16b, {v0.16b,v1.16b}, v3.16b \n"
...@@ -394,8 +392,7 @@ void OMITFP ScaleRowDown38_3_Box_NEON(const uint8* src_ptr, ...@@ -394,8 +392,7 @@ void OMITFP ScaleRowDown38_3_Box_NEON(const uint8* src_ptr,
"sqrdmulh v0.8h, v20.8h, v31.8h \n" "sqrdmulh v0.8h, v20.8h, v31.8h \n"
"sqrdmulh v1.8h, v21.8h, v31.8h \n" "sqrdmulh v1.8h, v21.8h, v31.8h \n"
// Align for table lookup, vtbl requires registers to // Align for table lookup, vtbl requires registers to be adjacent
// be adjacent
"tbl v3.16b, {v0.16b, v1.16b, v2.16b}, v30.16b \n" "tbl v3.16b, {v0.16b, v1.16b, v2.16b}, v30.16b \n"
"st1 {v3.8b}, [%1], #8 \n" "st1 {v3.8b}, [%1], #8 \n"
...@@ -776,8 +773,7 @@ void ScaleARGBRowDown2Box_NEON(const uint8* src_ptr, ...@@ -776,8 +773,7 @@ void ScaleARGBRowDown2Box_NEON(const uint8* src_ptr,
"uadalp v1.8h, v17.16b \n" // G 16 bytes -> 8 shorts. "uadalp v1.8h, v17.16b \n" // G 16 bytes -> 8 shorts.
"uadalp v2.8h, v18.16b \n" // R 16 bytes -> 8 shorts. "uadalp v2.8h, v18.16b \n" // R 16 bytes -> 8 shorts.
"uadalp v3.8h, v19.16b \n" // A 16 bytes -> 8 shorts. "uadalp v3.8h, v19.16b \n" // A 16 bytes -> 8 shorts.
"rshrn v0.8b, v0.8h, #2 \n" // downshift, round and "rshrn v0.8b, v0.8h, #2 \n" // round and pack
// pack
"rshrn v1.8b, v1.8h, #2 \n" "rshrn v1.8b, v1.8h, #2 \n"
"rshrn v2.8b, v2.8h, #2 \n" "rshrn v2.8b, v2.8h, #2 \n"
"rshrn v3.8b, v3.8h, #2 \n" "rshrn v3.8b, v3.8h, #2 \n"
...@@ -827,8 +823,7 @@ void ScaleARGBRowDownEvenBox_NEON(const uint8* src_argb, ...@@ -827,8 +823,7 @@ void ScaleARGBRowDownEvenBox_NEON(const uint8* src_argb,
asm volatile( asm volatile(
"add %1, %1, %0 \n" "add %1, %1, %0 \n"
"1: \n" "1: \n"
"ld1 {v0.8b}, [%0], %4 \n" // Read 4 2x2 blocks -> "ld1 {v0.8b}, [%0], %4 \n" // Read 4 2x2 -> 2x1
// 2x1
"ld1 {v1.8b}, [%1], %4 \n" "ld1 {v1.8b}, [%1], %4 \n"
"ld1 {v2.8b}, [%0], %4 \n" "ld1 {v2.8b}, [%0], %4 \n"
"ld1 {v3.8b}, [%1], %4 \n" "ld1 {v3.8b}, [%1], %4 \n"
...@@ -891,8 +886,7 @@ void ScaleARGBCols_NEON(uint8* dst_argb, ...@@ -891,8 +886,7 @@ void ScaleARGBCols_NEON(uint8* dst_argb,
LOAD1_DATA32_LANE(v1, 3) LOAD1_DATA32_LANE(v1, 3)
// clang-format on // clang-format on
"st1 {v0.4s, v1.4s}, [%0], #32 \n" // store pixels "st1 {v0.4s, v1.4s}, [%0], #32 \n" // store pixels
"subs %w2, %w2, #8 \n" // 8 processed per "subs %w2, %w2, #8 \n" // 8 processed per loop
// loop
"b.gt 1b \n" "b.gt 1b \n"
: "+r"(dst_argb), // %0 : "+r"(dst_argb), // %0
"+r"(src_argb), // %1 "+r"(src_argb), // %1
......
Markdown is supported
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