1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00

vp8: Use 2 registers for dst_stride and src_stride in neon bilin filter

Based on a patch by Ronald S. Bultje.

Signed-off-by: Martin Storsjö <martin@martin.st>
This commit is contained in:
Martin Storsjö 2014-02-05 14:47:32 +02:00
parent 5a0bccd281
commit 49ec551595

View File

@ -1576,18 +1576,18 @@ endconst
/* Bilinear MC */
function ff_put_vp8_bilin16_h_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
1:
subs r12, r12, #2
vld1.8 {d2-d4}, [r2], r1
vld1.8 {d2-d4}, [r2], r3
vext.8 q2, q1, q2, #1
vmull.u8 q8, d2, d1
vmlal.u8 q8, d4, d0
vld1.8 {d18-d20},[r2], r1
vld1.8 {d18-d20},[r2], r3
vmull.u8 q3, d3, d1
vmlal.u8 q3, d5, d0
vext.8 q10, q9, q10, #1
@ -1607,20 +1607,20 @@ function ff_put_vp8_bilin16_h_neon, export=1
endfunc
function ff_put_vp8_bilin16_v_neon, export=1
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #8] @ my
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
vld1.8 {q1}, [r2], r1
vld1.8 {q1}, [r2], r3
1:
subs r12, r12, #2
vld1.8 {q2}, [r2], r1
vld1.8 {q2}, [r2], r3
vmull.u8 q3, d2, d1
vmlal.u8 q3, d4, d0
vmull.u8 q8, d3, d1
vmlal.u8 q8, d5, d0
vld1.8 {q1}, [r2], r1
vld1.8 {q1}, [r2], r3
vmull.u8 q9, d4, d1
vmlal.u8 q9, d2, d0
vmull.u8 q10, d5, d1
@ -1637,17 +1637,17 @@ function ff_put_vp8_bilin16_v_neon, export=1
endfunc
function ff_put_vp8_bilin16_hv_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d2, r3
ldr r12, [sp, #8] @ my
vdup.8 d2, r12
rsb r12, r12, #8
vdup.8 d3, r12
ldr r12, [sp] @ h
vld1.8 {d4-d6}, [r2], r1
vld1.8 {d4-d6}, [r2], r3
vext.8 q3, q2, q3, #1
vmull.u8 q8, d4, d1
vmlal.u8 q8, d6, d0
@ -1657,11 +1657,11 @@ function ff_put_vp8_bilin16_hv_neon, export=1
vrshrn.u16 d5, q9, #3
1:
subs r12, r12, #2
vld1.8 {d18-d20},[r2], r1
vld1.8 {d18-d20},[r2], r3
vext.8 q10, q9, q10, #1
vmull.u8 q11, d18, d1
vmlal.u8 q11, d20, d0
vld1.8 {d26-d28},[r2], r1
vld1.8 {d26-d28},[r2], r3
vmull.u8 q12, d19, d1
vmlal.u8 q12, d21, d0
vext.8 q14, q13, q14, #1
@ -1693,18 +1693,18 @@ function ff_put_vp8_bilin16_hv_neon, export=1
endfunc
function ff_put_vp8_bilin8_h_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
1:
subs r12, r12, #2
vld1.8 {q1}, [r2], r1
vld1.8 {q1}, [r2], r3
vext.8 d3, d2, d3, #1
vmull.u8 q2, d2, d1
vmlal.u8 q2, d3, d0
vld1.8 {q3}, [r2], r1
vld1.8 {q3}, [r2], r3
vext.8 d7, d6, d7, #1
vmull.u8 q8, d6, d1
vmlal.u8 q8, d7, d0
@ -1718,18 +1718,18 @@ function ff_put_vp8_bilin8_h_neon, export=1
endfunc
function ff_put_vp8_bilin8_v_neon, export=1
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #8] @ my
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
vld1.8 {d2}, [r2], r1
vld1.8 {d2}, [r2], r3
1:
subs r12, r12, #2
vld1.8 {d3}, [r2], r1
vld1.8 {d3}, [r2], r3
vmull.u8 q2, d2, d1
vmlal.u8 q2, d3, d0
vld1.8 {d2}, [r2], r1
vld1.8 {d2}, [r2], r3
vmull.u8 q3, d3, d1
vmlal.u8 q3, d2, d0
vrshrn.u16 d4, q2, #3
@ -1742,28 +1742,28 @@ function ff_put_vp8_bilin8_v_neon, export=1
endfunc
function ff_put_vp8_bilin8_hv_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d2, r3
ldr r12, [sp, #8] @ my
vdup.8 d2, r12
rsb r12, r12, #8
vdup.8 d3, r12
ldr r12, [sp] @ h
vld1.8 {q2}, [r2], r1
vld1.8 {q2}, [r2], r3
vext.8 d5, d4, d5, #1
vmull.u8 q9, d4, d1
vmlal.u8 q9, d5, d0
vrshrn.u16 d22, q9, #3
1:
subs r12, r12, #2
vld1.8 {q3}, [r2], r1
vld1.8 {q3}, [r2], r3
vext.8 d7, d6, d7, #1
vmull.u8 q8, d6, d1
vmlal.u8 q8, d7, d0
vld1.8 {q2}, [r2], r1
vld1.8 {q2}, [r2], r3
vext.8 d5, d4, d5, #1
vmull.u8 q9, d4, d1
vmlal.u8 q9, d5, d0
@ -1783,16 +1783,16 @@ function ff_put_vp8_bilin8_hv_neon, export=1
endfunc
function ff_put_vp8_bilin4_h_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
1:
subs r12, r12, #2
vld1.8 {d2}, [r2], r1
vld1.8 {d2}, [r2], r3
vext.8 d3, d2, d3, #1
vld1.8 {d6}, [r2], r1
vld1.8 {d6}, [r2], r3
vext.8 d7, d6, d7, #1
vtrn.32 q1, q3
vmull.u8 q2, d2, d1
@ -1806,16 +1806,16 @@ function ff_put_vp8_bilin4_h_neon, export=1
endfunc
function ff_put_vp8_bilin4_v_neon, export=1
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #8] @ my
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r12, [sp] @ h
vld1.32 {d2[]}, [r2], r1
vld1.32 {d2[]}, [r2], r3
1:
vld1.32 {d3[]}, [r2]
vld1.32 {d2[1]}, [r2], r1
vld1.32 {d3[1]}, [r2], r1
vld1.32 {d2[1]}, [r2], r3
vld1.32 {d3[1]}, [r2], r3
vmull.u8 q2, d2, d1
vmlal.u8 q2, d3, d0
vtrn.32 d3, d2
@ -1829,26 +1829,26 @@ function ff_put_vp8_bilin4_v_neon, export=1
endfunc
function ff_put_vp8_bilin4_hv_neon, export=1
ldr r3, [sp, #4] @ mx
rsb r12, r3, #8
vdup.8 d0, r3
ldr r12, [sp, #4] @ mx
vdup.8 d0, r12
rsb r12, r12, #8
vdup.8 d1, r12
ldr r3, [sp, #8] @ my
rsb r12, r3, #8
vdup.8 d2, r3
ldr r12, [sp, #8] @ my
vdup.8 d2, r12
rsb r12, r12, #8
vdup.8 d3, r12
ldr r12, [sp] @ h
vld1.8 {d4}, [r2], r1
vld1.8 {d4}, [r2], r3
vext.8 d5, d4, d4, #1
vmull.u8 q9, d4, d1
vmlal.u8 q9, d5, d0
vrshrn.u16 d22, q9, #3
1:
subs r12, r12, #2
vld1.8 {d6}, [r2], r1
vld1.8 {d6}, [r2], r3
vext.8 d7, d6, d6, #1
vld1.8 {d4}, [r2], r1
vld1.8 {d4}, [r2], r3
vext.8 d5, d4, d4, #1
vtrn.32 q3, q2
vmull.u8 q8, d6, d1