%include "libavutil/x86/x86util.asm"
-%if ARCH_X86_64
SECTION_RODATA
pw_128: times 8 dw 128
SECTION .text
-INIT_XMM sse2
-cglobal blend_xor, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- pxor m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
-
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_or, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- por m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
-
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_and, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- pand m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
-
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_addition, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
+%macro BLEND_INIT 2
+%if ARCH_X86_64
+cglobal blend_%1, 6, 9, %2, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, end, x
+ mov widthd, dword widthm
+%else
+cglobal blend_%1, 5, 7, %2, top, top_linesize, bottom, bottom_linesize, dst, end, x
+%define dst_linesizeq r5mp
+%define widthq r6mp
+%endif
+ mov endd, dword r8m
add topq, widthq
add bottomq, widthq
add dstq, widthq
- sub endq, startq
+ sub endd, dword r7m ; start
neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- paddusb m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
+%endmacro
+%macro BLEND_END 0
add topq, top_linesizeq
add bottomq, bottom_linesizeq
add dstq, dst_linesizeq
sub endd, 1
jg .nextrow
REP_RET
+%endmacro
-cglobal blend_subtract, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
+%macro BLEND_SIMPLE 2
+BLEND_INIT %1, 2
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- psubusb m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
+ movu m0, [topq + xq]
+ movu m1, [bottomq + xq]
+ p%2 m0, m1
+ mova [dstq + xq], m0
+ add xq, mmsize
jl .loop
+BLEND_END
+%endmacro
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_difference128, 9, 11, 4, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+INIT_XMM sse2
+BLEND_SIMPLE xor, xor
+BLEND_SIMPLE or, or
+BLEND_SIMPLE and, and
+BLEND_SIMPLE addition, addusb
+BLEND_SIMPLE subtract, subusb
+BLEND_SIMPLE darken, minub
+BLEND_SIMPLE lighten, maxub
+
+BLEND_INIT difference128, 4
pxor m2, m2
mova m3, [pw_128]
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movh m0, [topq + x]
- movh m1, [bottomq + x]
+ movh m0, [topq + xq]
+ movh m1, [bottomq + xq]
punpcklbw m0, m2
punpcklbw m1, m2
paddw m0, m3
psubw m0, m1
packuswb m0, m0
- movh [dstq + x], m0
- add r10q, mmsize / 2
+ movh [dstq + xq], m0
+ add xq, mmsize / 2
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_average, 9, 11, 3, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+BLEND_INIT average, 3
pxor m2, m2
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movh m0, [topq + x]
- movh m1, [bottomq + x]
+ movh m0, [topq + xq]
+ movh m1, [bottomq + xq]
punpcklbw m0, m2
punpcklbw m1, m2
paddw m0, m1
psrlw m0, 1
packuswb m0, m0
- movh [dstq + x], m0
- add r10q, mmsize / 2
+ movh [dstq + xq], m0
+ add xq, mmsize / 2
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_addition128, 9, 11, 4, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+BLEND_INIT addition128, 4
pxor m2, m2
mova m3, [pw_128]
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movh m0, [topq + x]
- movh m1, [bottomq + x]
+ movh m0, [topq + xq]
+ movh m1, [bottomq + xq]
punpcklbw m0, m2
punpcklbw m1, m2
paddw m0, m1
psubw m0, m3
packuswb m0, m0
- movh [dstq + x], m0
- add r10q, mmsize / 2
+ movh [dstq + xq], m0
+ add xq, mmsize / 2
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_darken, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- pminub m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
-
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_hardmix, 9, 11, 5, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+BLEND_INIT hardmix, 5
mova m2, [pb_255]
mova m3, [pb_128]
mova m4, [pb_127]
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
+ movu m0, [topq + xq]
+ movu m1, [bottomq + xq]
pxor m1, m4
pxor m0, m3
pcmpgtb m1, m0
pxor m1, m2
- mova [dstq + x], m1
- add r10q, mmsize
+ mova [dstq + xq], m1
+ add xq, mmsize
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_lighten, 9, 11, 2, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
- neg widthq
-.nextrow:
- mov r10q, widthq
- %define x r10q
-
- .loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
- pmaxub m0, m1
- mova [dstq + x], m0
- add r10q, mmsize
- jl .loop
-
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_phoenix, 9, 11, 4, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+BLEND_INIT phoenix, 4
mova m3, [pb_255]
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movu m0, [topq + x]
- movu m1, [bottomq + x]
+ movu m0, [topq + xq]
+ movu m1, [bottomq + xq]
mova m2, m0
pminub m0, m1
pmaxub m1, m2
mova m2, m3
psubusb m2, m1
paddusb m2, m0
- mova [dstq + x], m2
- add r10q, mmsize
+ mova [dstq + xq], m2
+ add xq, mmsize
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-INIT_XMM ssse3
-cglobal blend_difference, 9, 11, 3, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+%macro BLEND_ABS 0
+BLEND_INIT difference, 3
pxor m2, m2
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movh m0, [topq + x]
- movh m1, [bottomq + x]
+ movh m0, [topq + xq]
+ movh m1, [bottomq + xq]
punpcklbw m0, m2
punpcklbw m1, m2
psubw m0, m1
- pabsw m0, m0
+ ABS1 m0, m1
packuswb m0, m0
- movh [dstq + x], m0
- add r10q, mmsize / 2
+ movh [dstq + xq], m0
+ add xq, mmsize / 2
jl .loop
+BLEND_END
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-cglobal blend_negation, 9, 11, 5, 0, top, top_linesize, bottom, bottom_linesize, dst, dst_linesize, width, start, end
- add topq, widthq
- add bottomq, widthq
- add dstq, widthq
- sub endq, startq
+BLEND_INIT negation, 5
pxor m2, m2
mova m4, [pw_255]
- neg widthq
.nextrow:
- mov r10q, widthq
- %define x r10q
+ mov xq, widthq
.loop:
- movh m0, [topq + x]
- movh m1, [bottomq + x]
+ movh m0, [topq + xq]
+ movh m1, [bottomq + xq]
punpcklbw m0, m2
punpcklbw m1, m2
mova m3, m4
psubw m3, m0
psubw m3, m1
- pabsw m3, m3
+ ABS1 m3, m1
mova m0, m4
psubw m0, m3
packuswb m0, m0
- movh [dstq + x], m0
- add r10q, mmsize / 2
+ movh [dstq + xq], m0
+ add xq, mmsize / 2
jl .loop
+BLEND_END
+%endmacro
- add topq, top_linesizeq
- add bottomq, bottom_linesizeq
- add dstq, dst_linesizeq
- sub endd, 1
- jg .nextrow
-REP_RET
-
-%endif
+INIT_XMM sse2
+BLEND_ABS
+INIT_XMM ssse3
+BLEND_ABS
projects / ffmpeg / blobdiff