MrBesen
/
ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

swresample/arm: cosmetic fixes

This commit is contained in:
Matthieu Bouron 2017-01-09 17:22:40 +01:00
parent 0265aec565
commit e109c54a69
2 changed files with 22 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
libswresample/arm/resample.S
View File

@ -22,9 +22,9 @@
function ff_resample_common_apply_filter_x4_float_neon, export=1
vmov.f32 q0, #0.0 @ accumulator
1: vld1.32 {q1}, [r1]! @ src
vld1.32 {q2}, [r2]! @ filter
vmla.f32 q0, q1, q2 @ accumulator += src + {0..3} * filter + {0..3}
1: vld1.32 {q1}, [r1]! @ src[0..3]
vld1.32 {q2}, [r2]! @ filter[0..3]
vmla.f32 q0, q1, q2 @ accumulator += src[0..3] * filter[0..3]
subs r3, #4 @ filter_length -= 4
bgt 1b @ loop until filter_length
vpadd.f32 d0, d0, d1 @ pair adding of the 4x32-bit accumulated values
@ -35,12 +35,12 @@ endfunc
function ff_resample_common_apply_filter_x8_float_neon, export=1
vmov.f32 q0, #0.0 @ accumulator
1: vld1.32 {q1}, [r1]! @ src
vld1.32 {q2}, [r2]! @ filter
vld1.32 {q8}, [r1]! @ src
vld1.32 {q9}, [r2]! @ filter
vmla.f32 q0, q1, q2 @ accumulator += src + {0..3} * filter + {0..3}
vmla.f32 q0, q8, q9 @ accumulator += src + {4..7} * filter + {4..7}
1: vld1.32 {q1}, [r1]! @ src[0..3]
vld1.32 {q2}, [r2]! @ filter[0..3]
vld1.32 {q8}, [r1]! @ src[4..7]
vld1.32 {q9}, [r2]! @ filter[4..7]
vmla.f32 q0, q1, q2 @ accumulator += src[0..3] * filter[0..3]
vmla.f32 q0, q8, q9 @ accumulator += src[4..7] * filter[4..7]
subs r3, #8 @ filter_length -= 8
bgt 1b @ loop until filter_length
vpadd.f32 d0, d0, d1 @ pair adding of the 4x32-bit accumulated values
@ -51,9 +51,9 @@ endfunc
function ff_resample_common_apply_filter_x4_s16_neon, export=1
vmov.s32 q0, #0 @ accumulator
1: vld1.16 {d2}, [r1]! @ src
vld1.16 {d4}, [r2]! @ filter
vmlal.s16 q0, d2, d4 @ accumulator += src + {0..3} * filter + {0..3}
1: vld1.16 {d2}, [r1]! @ src[0..3]
vld1.16 {d4}, [r2]! @ filter[0..3]
vmlal.s16 q0, d2, d4 @ accumulator += src[0..3] * filter[0..3]
subs r3, #4 @ filter_length -= 4
bgt 1b @ loop until filter_length
vpadd.s32 d0, d0, d1 @ pair adding of the 4x32-bit accumulated values
@ -64,10 +64,10 @@ endfunc
function ff_resample_common_apply_filter_x8_s16_neon, export=1
vmov.s32 q0, #0 @ accumulator
1: vld1.16 {q1}, [r1]! @ src
vld1.16 {q2}, [r2]! @ filter
vmlal.s16 q0, d2, d4 @ accumulator += src + {0..3} * filter + {0..3}
vmlal.s16 q0, d3, d5 @ accumulator += src + {4..7} * filter + {4..7}
1: vld1.16 {q1}, [r1]! @ src[0..7]
vld1.16 {q2}, [r2]! @ filter[0..7]
vmlal.s16 q0, d2, d4 @ accumulator += src[0..3] * filter[0..3]
vmlal.s16 q0, d3, d5 @ accumulator += src[4..7] * filter[4..7]
subs r3, #8 @ filter_length -= 8
bgt 1b @ loop until filter_length
vpadd.s32 d0, d0, d1 @ pair adding of the 4x32-bit accumulated values