* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
-#include "asm.S"
+#include "libavutil/arm/asm.S"
/* chroma_mc8(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y) */
-.macro h264_chroma_mc8 type
-function ff_\type\()_h264_chroma_mc8_neon, export=1
+.macro h264_chroma_mc8 type, codec=h264
+function ff_\type\()_\codec\()_chroma_mc8_neon, export=1
push {r4-r7, lr}
- ldrd r4, [sp, #20]
+ ldrd r4, r5, [sp, #20]
.ifc \type,avg
mov lr, r0
.endif
pld [r1]
pld [r1, r2]
+ .ifc \codec,rv40
+ movrel r6, rv40bias
+ lsr r7, r5, #1
+ add r6, r6, r7, lsl #3
+ lsr r7, r4, #1
+ add r6, r6, r7, lsl #1
+ vld1.16 {d22[],d23[]}, [r6,:16]
+ .endif
+ .ifc \codec,vc1
+ vmov.u16 q11, #28
+ .endif
+
A muls r7, r4, r5
T mul r7, r4, r5
T cmp r7, #0
beq 2f
- add r5, r1, r2
-
vdup.8 d0, r4
- lsl r4, r2, #1
vdup.8 d1, r12
- vld1.8 {d4, d5}, [r1], r4
+ vld1.8 {d4, d5}, [r1], r2
vdup.8 d2, r6
- vld1.8 {d6, d7}, [r5], r4
vdup.8 d3, r7
-
vext.8 d5, d4, d5, #1
- vext.8 d7, d6, d7, #1
-1: pld [r5]
+1: vld1.8 {d6, d7}, [r1], r2
vmull.u8 q8, d4, d0
vmlal.u8 q8, d5, d1
- vld1.8 {d4, d5}, [r1], r4
+ vext.8 d7, d6, d7, #1
+ vld1.8 {d4, d5}, [r1], r2
vmlal.u8 q8, d6, d2
+ pld [r1]
vext.8 d5, d4, d5, #1
vmlal.u8 q8, d7, d3
vmull.u8 q9, d6, d0
vmlal.u8 q9, d7, d1
vmlal.u8 q9, d4, d2
vmlal.u8 q9, d5, d3
+ pld [r1, r2]
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
- vld1.8 {d6, d7}, [r5], r4
- pld [r1]
vrshrn.u16 d17, q9, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vadd.u16 q9, q9, q11
+ vshrn.u16 d16, q8, #6
+ vshrn.u16 d17, q9, #6
+ .endif
.ifc \type,avg
vld1.8 {d20}, [lr,:64], r2
vld1.8 {d21}, [lr,:64], r2
vrhadd.u8 q8, q8, q10
.endif
- vext.8 d7, d6, d7, #1
vst1.8 {d16}, [r0,:64], r2
vst1.8 {d17}, [r0,:64], r2
bgt 1b
beq 4f
- add r5, r1, r2
- lsl r4, r2, #1
- vld1.8 {d4}, [r1], r4
- vld1.8 {d6}, [r5], r4
+ vld1.8 {d4}, [r1], r2
-3: pld [r5]
+3: vld1.8 {d6}, [r1], r2
vmull.u8 q8, d4, d0
vmlal.u8 q8, d6, d1
- vld1.8 {d4}, [r1], r4
+ vld1.8 {d4}, [r1], r2
vmull.u8 q9, d6, d0
vmlal.u8 q9, d4, d1
- vld1.8 {d6}, [r5], r4
+ pld [r1]
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
vrshrn.u16 d17, q9, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vadd.u16 q9, q9, q11
+ vshrn.u16 d16, q8, #6
+ vshrn.u16 d17, q9, #6
+ .endif
+ pld [r1, r2]
.ifc \type,avg
vld1.8 {d20}, [lr,:64], r2
vld1.8 {d21}, [lr,:64], r2
vrhadd.u8 q8, q8, q10
.endif
subs r3, r3, #2
- pld [r1]
vst1.8 {d16}, [r0,:64], r2
vst1.8 {d17}, [r0,:64], r2
bgt 3b
vld1.8 {d6, d7}, [r1], r2
vext.8 d5, d4, d5, #1
vext.8 d7, d6, d7, #1
-
-5: pld [r1]
+ pld [r1]
subs r3, r3, #2
vmull.u8 q8, d4, d0
vmlal.u8 q8, d5, d1
- vld1.8 {d4, d5}, [r1], r2
vmull.u8 q9, d6, d0
vmlal.u8 q9, d7, d1
- pld [r1]
- vext.8 d5, d4, d5, #1
+ pld [r1, r2]
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
vrshrn.u16 d17, q9, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vadd.u16 q9, q9, q11
+ vshrn.u16 d16, q8, #6
+ vshrn.u16 d17, q9, #6
+ .endif
.ifc \type,avg
vld1.8 {d20}, [lr,:64], r2
vld1.8 {d21}, [lr,:64], r2
vrhadd.u8 q8, q8, q10
.endif
- vld1.8 {d6, d7}, [r1], r2
- vext.8 d7, d6, d7, #1
vst1.8 {d16}, [r0,:64], r2
vst1.8 {d17}, [r0,:64], r2
- bgt 5b
+ bgt 4b
pop {r4-r7, pc}
endfunc
.endm
/* chroma_mc4(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y) */
-.macro h264_chroma_mc4 type
-function ff_\type\()_h264_chroma_mc4_neon, export=1
+.macro h264_chroma_mc4 type, codec=h264
+function ff_\type\()_\codec\()_chroma_mc4_neon, export=1
push {r4-r7, lr}
- ldrd r4, [sp, #20]
+ ldrd r4, r5, [sp, #20]
.ifc \type,avg
mov lr, r0
.endif
pld [r1]
pld [r1, r2]
+ .ifc \codec,rv40
+ movrel r6, rv40bias
+ lsr r7, r5, #1
+ add r6, r6, r7, lsl #3
+ lsr r7, r4, #1
+ add r6, r6, r7, lsl #1
+ vld1.16 {d22[],d23[]}, [r6,:16]
+ .endif
+ .ifc \codec,vc1
+ vmov.u16 q11, #28
+ .endif
+
A muls r7, r4, r5
T mul r7, r4, r5
T cmp r7, #0
beq 2f
- add r5, r1, r2
-
vdup.8 d0, r4
- lsl r4, r2, #1
vdup.8 d1, r12
- vld1.8 {d4}, [r1], r4
+ vld1.8 {d4}, [r1], r2
vdup.8 d2, r6
- vld1.8 {d6}, [r5], r4
vdup.8 d3, r7
vext.8 d5, d4, d5, #1
- vext.8 d7, d6, d7, #1
vtrn.32 d4, d5
- vtrn.32 d6, d7
vtrn.32 d0, d1
vtrn.32 d2, d3
-1: pld [r5]
+1: vld1.8 {d6}, [r1], r2
+ vext.8 d7, d6, d7, #1
+ vtrn.32 d6, d7
vmull.u8 q8, d4, d0
vmlal.u8 q8, d6, d2
- vld1.8 {d4}, [r1], r4
+ vld1.8 {d4}, [r1], r2
vext.8 d5, d4, d5, #1
vtrn.32 d4, d5
+ pld [r1]
vmull.u8 q9, d6, d0
vmlal.u8 q9, d4, d2
- vld1.8 {d6}, [r5], r4
vadd.i16 d16, d16, d17
vadd.i16 d17, d18, d19
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vshrn.u16 d16, q8, #6
+ .endif
subs r3, r3, #2
- pld [r1]
+ pld [r1, r2]
.ifc \type,avg
vld1.32 {d20[0]}, [lr,:32], r2
vld1.32 {d20[1]}, [lr,:32], r2
vrhadd.u8 d16, d16, d20
.endif
- vext.8 d7, d6, d7, #1
- vtrn.32 d6, d7
vst1.32 {d16[0]}, [r0,:32], r2
vst1.32 {d16[1]}, [r0,:32], r2
bgt 1b
beq 4f
vext.32 d1, d0, d1, #1
- add r5, r1, r2
- lsl r4, r2, #1
- vld1.32 {d4[0]}, [r1], r4
- vld1.32 {d4[1]}, [r5], r4
+ vld1.32 {d4[0]}, [r1], r2
-3: pld [r5]
+3: vld1.32 {d4[1]}, [r1], r2
vmull.u8 q8, d4, d0
- vld1.32 {d4[0]}, [r1], r4
+ vld1.32 {d4[0]}, [r1], r2
vmull.u8 q9, d4, d1
- vld1.32 {d4[1]}, [r5], r4
vadd.i16 d16, d16, d17
vadd.i16 d17, d18, d19
+ pld [r1]
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vshrn.u16 d16, q8, #6
+ .endif
.ifc \type,avg
vld1.32 {d20[0]}, [lr,:32], r2
vld1.32 {d20[1]}, [lr,:32], r2
vrhadd.u8 d16, d16, d20
.endif
subs r3, r3, #2
- pld [r1]
+ pld [r1, r2]
vst1.32 {d16[0]}, [r0,:32], r2
vst1.32 {d16[1]}, [r0,:32], r2
bgt 3b
vext.8 d7, d6, d7, #1
vtrn.32 d4, d5
vtrn.32 d6, d7
-
-5: vmull.u8 q8, d4, d0
+ vmull.u8 q8, d4, d0
vmull.u8 q9, d6, d0
subs r3, r3, #2
- vld1.8 {d4}, [r1], r2
- vext.8 d5, d4, d5, #1
- vtrn.32 d4, d5
vadd.i16 d16, d16, d17
vadd.i16 d17, d18, d19
pld [r1]
+ .ifc \codec,h264
vrshrn.u16 d16, q8, #6
+ .else
+ vadd.u16 q8, q8, q11
+ vshrn.u16 d16, q8, #6
+ .endif
.ifc \type,avg
vld1.32 {d20[0]}, [lr,:32], r2
vld1.32 {d20[1]}, [lr,:32], r2
vrhadd.u8 d16, d16, d20
.endif
- vld1.8 {d6}, [r1], r2
- vext.8 d7, d6, d7, #1
- vtrn.32 d6, d7
pld [r1]
vst1.32 {d16[0]}, [r0,:32], r2
vst1.32 {d16[1]}, [r0,:32], r2
- bgt 5b
+ bgt 4b
pop {r4-r7, pc}
endfunc
h264_chroma_mc4 avg
h264_chroma_mc2 put
h264_chroma_mc2 avg
+
+#if CONFIG_RV40_DECODER
+const rv40bias
+ .short 0, 16, 32, 16
+ .short 32, 28, 32, 28
+ .short 0, 32, 16, 32
+ .short 32, 28, 32, 28
+endconst
+
+ h264_chroma_mc8 put, rv40
+ h264_chroma_mc8 avg, rv40
+ h264_chroma_mc4 put, rv40
+ h264_chroma_mc4 avg, rv40
+#endif
+
+#if CONFIG_VC1_DECODER
+ h264_chroma_mc8 put, vc1
+ h264_chroma_mc8 avg, vc1
+ h264_chroma_mc4 put, vc1
+ h264_chroma_mc4 avg, vc1
+#endif
projects / ffmpeg / blobdiff