SkJumper: scales and lerps
Change-Id: I6057ba3e9243641fecbc6b78f6f83ee3265ad3d4
Reviewed-on: https://skia-review.googlesource.com/8941
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
diff --git a/src/jumper/SkJumper.cpp b/src/jumper/SkJumper.cpp
index b1e9317..90db9a9 100644
--- a/src/jumper/SkJumper.cpp
+++ b/src/jumper/SkJumper.cpp
@@ -64,8 +64,11 @@
M(unpremul) \
M(from_srgb) \
M(to_srgb) \
+ M(scale_1_float) \
M(scale_u8) \
+ M(lerp_1_float) \
M(lerp_u8) \
+ M(lerp_565) \
M(load_tables) \
M(load_565) \
M(store_565) \
diff --git a/src/jumper/SkJumper_generated.S b/src/jumper/SkJumper_generated.S
index ab8b8b2..8b4106c 100644
--- a/src/jumper/SkJumper_generated.S
+++ b/src/jumper/SkJumper_generated.S
@@ -350,6 +350,16 @@
.long 0x91004021 // add x1, x1, #0x10
.long 0xd61f0060 // br x3
+.globl _sk_scale_1_float_aarch64
+_sk_scale_1_float_aarch64:
+ .long 0xa8c10c28 // ldp x8, x3, [x1],#16
+ .long 0xbd400110 // ldr s16, [x8]
+ .long 0x4f909000 // fmul v0.4s, v0.4s, v16.s[0]
+ .long 0x4f909021 // fmul v1.4s, v1.4s, v16.s[0]
+ .long 0x4f909042 // fmul v2.4s, v2.4s, v16.s[0]
+ .long 0x4f909063 // fmul v3.4s, v3.4s, v16.s[0]
+ .long 0xd61f0060 // br x3
+
.globl _sk_scale_u8_aarch64
_sk_scale_u8_aarch64:
.long 0xa8c10c28 // ldp x8, x3, [x1],#16
@@ -374,6 +384,24 @@
.long 0x6e23de03 // fmul v3.4s, v16.4s, v3.4s
.long 0xd61f0060 // br x3
+.globl _sk_lerp_1_float_aarch64
+_sk_lerp_1_float_aarch64:
+ .long 0xa8c10c28 // ldp x8, x3, [x1],#16
+ .long 0x4ea4d411 // fsub v17.4s, v0.4s, v4.4s
+ .long 0x4ea41c80 // mov v0.16b, v4.16b
+ .long 0x4ea5d432 // fsub v18.4s, v1.4s, v5.4s
+ .long 0xbd400110 // ldr s16, [x8]
+ .long 0x4ea51ca1 // mov v1.16b, v5.16b
+ .long 0x4f901220 // fmla v0.4s, v17.4s, v16.s[0]
+ .long 0x4ea6d451 // fsub v17.4s, v2.4s, v6.4s
+ .long 0x4f901241 // fmla v1.4s, v18.4s, v16.s[0]
+ .long 0x4ea61cc2 // mov v2.16b, v6.16b
+ .long 0x4ea7d472 // fsub v18.4s, v3.4s, v7.4s
+ .long 0x4ea71ce3 // mov v3.16b, v7.16b
+ .long 0x4f901222 // fmla v2.4s, v17.4s, v16.s[0]
+ .long 0x4f901243 // fmla v3.4s, v18.4s, v16.s[0]
+ .long 0xd61f0060 // br x3
+
.globl _sk_lerp_u8_aarch64
_sk_lerp_u8_aarch64:
.long 0xa8c10c28 // ldp x8, x3, [x1],#16
@@ -406,6 +434,42 @@
.long 0x4e31ce03 // fmla v3.4s, v16.4s, v17.4s
.long 0xd61f0060 // br x3
+.globl _sk_lerp_565_aarch64
+_sk_lerp_565_aarch64:
+ .long 0xa8c10c28 // ldp x8, x3, [x1],#16
+ .long 0xd37ff809 // lsl x9, x0, #1
+ .long 0x4ea4d413 // fsub v19.4s, v0.4s, v4.4s
+ .long 0x4ea41c80 // mov v0.16b, v4.16b
+ .long 0xf9400108 // ldr x8, [x8]
+ .long 0xfc696903 // ldr d3, [x8,x9]
+ .long 0x9101a048 // add x8, x2, #0x68
+ .long 0x4d40c910 // ld1r {v16.4s}, [x8]
+ .long 0x2d4ec851 // ldp s17, s18, [x2,#116]
+ .long 0x2f10a463 // uxtl v3.4s, v3.4h
+ .long 0x9101b048 // add x8, x2, #0x6c
+ .long 0x4e231e10 // and v16.16b, v16.16b, v3.16b
+ .long 0x4e21da10 // scvtf v16.4s, v16.4s
+ .long 0x4f919210 // fmul v16.4s, v16.4s, v17.s[0]
+ .long 0x4d40c911 // ld1r {v17.4s}, [x8]
+ .long 0x9101c048 // add x8, x2, #0x70
+ .long 0x4e33ce00 // fmla v0.4s, v16.4s, v19.4s
+ .long 0x4ea5d430 // fsub v16.4s, v1.4s, v5.4s
+ .long 0x4e231e31 // and v17.16b, v17.16b, v3.16b
+ .long 0x4e21da31 // scvtf v17.4s, v17.4s
+ .long 0x4f929231 // fmul v17.4s, v17.4s, v18.s[0]
+ .long 0x4d40c912 // ld1r {v18.4s}, [x8]
+ .long 0x4ea51ca1 // mov v1.16b, v5.16b
+ .long 0x4e30ce21 // fmla v1.4s, v17.4s, v16.4s
+ .long 0xbd407c50 // ldr s16, [x2,#124]
+ .long 0x4e231e52 // and v18.16b, v18.16b, v3.16b
+ .long 0x4d40c843 // ld1r {v3.4s}, [x2]
+ .long 0x4e21da52 // scvtf v18.4s, v18.4s
+ .long 0x4ea6d451 // fsub v17.4s, v2.4s, v6.4s
+ .long 0x4ea61cc2 // mov v2.16b, v6.16b
+ .long 0x4f909250 // fmul v16.4s, v18.4s, v16.s[0]
+ .long 0x4e31ce02 // fmla v2.4s, v16.4s, v17.4s
+ .long 0xd61f0060 // br x3
+
.globl _sk_load_tables_aarch64
_sk_load_tables_aarch64:
.long 0xa8c10c28 // ldp x8, x3, [x1],#16
@@ -1026,6 +1090,20 @@
.long 0xecbd8b02 // vpop {d8}
.long 0xe12fff13 // bx r3
+.globl _sk_scale_1_float_vfp4
+_sk_scale_1_float_vfp4:
+ .long 0xed2d8b02 // vpush {d8}
+ .long 0xe5913000 // ldr r3, [r1]
+ .long 0xe591c004 // ldr ip, [r1, #4]
+ .long 0xe2811008 // add r1, r1, #8
+ .long 0xed938a00 // vldr s16, [r3]
+ .long 0xf2a00948 // vmul.f32 d0, d0, d8[0]
+ .long 0xf2a11948 // vmul.f32 d1, d1, d8[0]
+ .long 0xf2a22948 // vmul.f32 d2, d2, d8[0]
+ .long 0xf2a33948 // vmul.f32 d3, d3, d8[0]
+ .long 0xecbd8b02 // vpop {d8}
+ .long 0xe12fff1c // bx ip
+
.globl _sk_scale_u8_vfp4
_sk_scale_u8_vfp4:
.long 0xed2d8b02 // vpush {d8}
@@ -1052,6 +1130,26 @@
.long 0xecbd8b02 // vpop {d8}
.long 0xe12fff1c // bx ip
+.globl _sk_lerp_1_float_vfp4
+_sk_lerp_1_float_vfp4:
+ .long 0xe5913000 // ldr r3, [r1]
+ .long 0xf2600d04 // vsub.f32 d16, d0, d4
+ .long 0xf2611d05 // vsub.f32 d17, d1, d5
+ .long 0xe591c004 // ldr ip, [r1, #4]
+ .long 0xf2622d06 // vsub.f32 d18, d2, d6
+ .long 0xe2811008 // add r1, r1, #8
+ .long 0xf2633d07 // vsub.f32 d19, d3, d7
+ .long 0xf4e34c9f // vld1.32 {d20[]}, [r3 :32]
+ .long 0xf2240114 // vorr d0, d4, d4
+ .long 0xf2251115 // vorr d1, d5, d5
+ .long 0xf2262116 // vorr d2, d6, d6
+ .long 0xf2273117 // vorr d3, d7, d7
+ .long 0xf2000cb4 // vfma.f32 d0, d16, d20
+ .long 0xf2011cb4 // vfma.f32 d1, d17, d20
+ .long 0xf2022cb4 // vfma.f32 d2, d18, d20
+ .long 0xf2033cb4 // vfma.f32 d3, d19, d20
+ .long 0xe12fff1c // bx ip
+
.globl _sk_lerp_u8_vfp4
_sk_lerp_u8_vfp4:
.long 0xed2d8b02 // vpush {d8}
@@ -1086,6 +1184,51 @@
.long 0xecbd8b02 // vpop {d8}
.long 0xe12fff1c // bx ip
+.globl _sk_lerp_565_vfp4
+_sk_lerp_565_vfp4:
+ .long 0xed2d8b04 // vpush {d8-d9}
+ .long 0xe24dd008 // sub sp, sp, #8
+ .long 0xe5913000 // ldr r3, [r1]
+ .long 0xf2603d04 // vsub.f32 d19, d0, d4
+ .long 0xe591c004 // ldr ip, [r1, #4]
+ .long 0xf2240114 // vorr d0, d4, d4
+ .long 0xe2811008 // add r1, r1, #8
+ .long 0xe5933000 // ldr r3, [r3]
+ .long 0xe7933080 // ldr r3, [r3, r0, lsl #1]
+ .long 0xe58d3004 // str r3, [sp, #4]
+ .long 0xe28d3004 // add r3, sp, #4
+ .long 0xed923a1d // vldr s6, [r2, #116]
+ .long 0xf4e3083f // vld1.32 {d16[0]}, [r3 :32]
+ .long 0xe282306c // add r3, r2, #108
+ .long 0xf4e31c9f // vld1.32 {d17[]}, [r3 :32]
+ .long 0xe2823068 // add r3, r2, #104
+ .long 0xf3d04a30 // vmovl.u16 q10, d16
+ .long 0xf4e32c9f // vld1.32 {d18[]}, [r3 :32]
+ .long 0xe2823070 // add r3, r2, #112
+ .long 0xf24201b4 // vand d16, d18, d20
+ .long 0xf4e32c9f // vld1.32 {d18[]}, [r3 :32]
+ .long 0xf24221b4 // vand d18, d18, d20
+ .long 0xf24111b4 // vand d17, d17, d20
+ .long 0xf3fb0620 // vcvt.f32.s32 d16, d16
+ .long 0xed928a1e // vldr s16, [r2, #120]
+ .long 0xf3fb1621 // vcvt.f32.s32 d17, d17
+ .long 0xed929a1f // vldr s18, [r2, #124]
+ .long 0xf3fb2622 // vcvt.f32.s32 d18, d18
+ .long 0xf2614d05 // vsub.f32 d20, d1, d5
+ .long 0xf2e009c3 // vmul.f32 d16, d16, d3[0]
+ .long 0xf4a23c9f // vld1.32 {d3[]}, [r2 :32]
+ .long 0xf2625d06 // vsub.f32 d21, d2, d6
+ .long 0xf2e119c8 // vmul.f32 d17, d17, d8[0]
+ .long 0xf2e229c9 // vmul.f32 d18, d18, d9[0]
+ .long 0xf2251115 // vorr d1, d5, d5
+ .long 0xf2262116 // vorr d2, d6, d6
+ .long 0xf2030cb0 // vfma.f32 d0, d19, d16
+ .long 0xf2041cb1 // vfma.f32 d1, d20, d17
+ .long 0xf2052cb2 // vfma.f32 d2, d21, d18
+ .long 0xe28dd008 // add sp, sp, #8
+ .long 0xecbd8b04 // vpop {d8-d9}
+ .long 0xe12fff1c // bx ip
+
.globl _sk_load_tables_vfp4
_sk_load_tables_vfp4:
.long 0xe92d48f0 // push {r4, r5, r6, r7, fp, lr}
@@ -1715,6 +1858,17 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_scale_1_float_hsw
+_sk_scale_1_float_hsw:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 196,98,125,24,0 // vbroadcastss (%rax),%ymm8
+ .byte 197,188,89,192 // vmulps %ymm0,%ymm8,%ymm0
+ .byte 197,188,89,201 // vmulps %ymm1,%ymm8,%ymm1
+ .byte 197,188,89,210 // vmulps %ymm2,%ymm8,%ymm2
+ .byte 197,188,89,219 // vmulps %ymm3,%ymm8,%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_scale_u8_hsw
_sk_scale_u8_hsw:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -1730,6 +1884,21 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_1_float_hsw
+_sk_lerp_1_float_hsw:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 196,98,125,24,0 // vbroadcastss (%rax),%ymm8
+ .byte 197,252,92,196 // vsubps %ymm4,%ymm0,%ymm0
+ .byte 196,226,61,168,196 // vfmadd213ps %ymm4,%ymm8,%ymm0
+ .byte 197,244,92,205 // vsubps %ymm5,%ymm1,%ymm1
+ .byte 196,226,61,168,205 // vfmadd213ps %ymm5,%ymm8,%ymm1
+ .byte 197,236,92,214 // vsubps %ymm6,%ymm2,%ymm2
+ .byte 196,226,61,168,214 // vfmadd213ps %ymm6,%ymm8,%ymm2
+ .byte 197,228,92,223 // vsubps %ymm7,%ymm3,%ymm3
+ .byte 196,226,61,168,223 // vfmadd213ps %ymm7,%ymm8,%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_lerp_u8_hsw
_sk_lerp_u8_hsw:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -1749,6 +1918,36 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_565_hsw
+_sk_lerp_565_hsw:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 72,139,0 // mov (%rax),%rax
+ .byte 196,226,125,51,28,120 // vpmovzxwd (%rax,%rdi,2),%ymm3
+ .byte 196,98,125,88,66,104 // vpbroadcastd 0x68(%rdx),%ymm8
+ .byte 197,61,219,195 // vpand %ymm3,%ymm8,%ymm8
+ .byte 196,65,124,91,192 // vcvtdq2ps %ymm8,%ymm8
+ .byte 196,98,125,24,74,116 // vbroadcastss 0x74(%rdx),%ymm9
+ .byte 196,65,52,89,192 // vmulps %ymm8,%ymm9,%ymm8
+ .byte 196,98,125,88,74,108 // vpbroadcastd 0x6c(%rdx),%ymm9
+ .byte 197,53,219,203 // vpand %ymm3,%ymm9,%ymm9
+ .byte 196,65,124,91,201 // vcvtdq2ps %ymm9,%ymm9
+ .byte 196,98,125,24,82,120 // vbroadcastss 0x78(%rdx),%ymm10
+ .byte 196,65,44,89,201 // vmulps %ymm9,%ymm10,%ymm9
+ .byte 196,98,125,88,82,112 // vpbroadcastd 0x70(%rdx),%ymm10
+ .byte 197,173,219,219 // vpand %ymm3,%ymm10,%ymm3
+ .byte 197,252,91,219 // vcvtdq2ps %ymm3,%ymm3
+ .byte 196,98,125,24,82,124 // vbroadcastss 0x7c(%rdx),%ymm10
+ .byte 197,172,89,219 // vmulps %ymm3,%ymm10,%ymm3
+ .byte 197,252,92,196 // vsubps %ymm4,%ymm0,%ymm0
+ .byte 196,226,61,168,196 // vfmadd213ps %ymm4,%ymm8,%ymm0
+ .byte 197,244,92,205 // vsubps %ymm5,%ymm1,%ymm1
+ .byte 196,226,53,168,205 // vfmadd213ps %ymm5,%ymm9,%ymm1
+ .byte 197,236,92,214 // vsubps %ymm6,%ymm2,%ymm2
+ .byte 196,226,101,168,214 // vfmadd213ps %ymm6,%ymm3,%ymm2
+ .byte 196,226,125,24,26 // vbroadcastss (%rdx),%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_load_tables_hsw
_sk_load_tables_hsw:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -2336,6 +2535,17 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_scale_1_float_avx
+_sk_scale_1_float_avx:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 196,98,125,24,0 // vbroadcastss (%rax),%ymm8
+ .byte 197,188,89,192 // vmulps %ymm0,%ymm8,%ymm0
+ .byte 197,188,89,201 // vmulps %ymm1,%ymm8,%ymm1
+ .byte 197,188,89,210 // vmulps %ymm2,%ymm8,%ymm2
+ .byte 197,188,89,219 // vmulps %ymm3,%ymm8,%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_scale_u8_avx
_sk_scale_u8_avx:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -2353,6 +2563,25 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_1_float_avx
+_sk_lerp_1_float_avx:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 196,98,125,24,0 // vbroadcastss (%rax),%ymm8
+ .byte 197,252,92,196 // vsubps %ymm4,%ymm0,%ymm0
+ .byte 196,193,124,89,192 // vmulps %ymm8,%ymm0,%ymm0
+ .byte 197,252,88,196 // vaddps %ymm4,%ymm0,%ymm0
+ .byte 197,244,92,205 // vsubps %ymm5,%ymm1,%ymm1
+ .byte 196,193,116,89,200 // vmulps %ymm8,%ymm1,%ymm1
+ .byte 197,244,88,205 // vaddps %ymm5,%ymm1,%ymm1
+ .byte 197,236,92,214 // vsubps %ymm6,%ymm2,%ymm2
+ .byte 196,193,108,89,208 // vmulps %ymm8,%ymm2,%ymm2
+ .byte 197,236,88,214 // vaddps %ymm6,%ymm2,%ymm2
+ .byte 197,228,92,223 // vsubps %ymm7,%ymm3,%ymm3
+ .byte 196,193,100,89,216 // vmulps %ymm8,%ymm3,%ymm3
+ .byte 197,228,88,223 // vaddps %ymm7,%ymm3,%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_lerp_u8_avx
_sk_lerp_u8_avx:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -2378,6 +2607,47 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_565_avx
+_sk_lerp_565_avx:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 72,139,0 // mov (%rax),%rax
+ .byte 196,226,121,51,92,120,8 // vpmovzxwd 0x8(%rax,%rdi,2),%xmm3
+ .byte 196,98,121,51,4,120 // vpmovzxwd (%rax,%rdi,2),%xmm8
+ .byte 196,99,61,24,195,1 // vinsertf128 $0x1,%xmm3,%ymm8,%ymm8
+ .byte 197,249,110,90,104 // vmovd 0x68(%rdx),%xmm3
+ .byte 196,227,121,4,219,0 // vpermilps $0x0,%xmm3,%xmm3
+ .byte 196,227,101,24,219,1 // vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ .byte 196,193,100,84,216 // vandps %ymm8,%ymm3,%ymm3
+ .byte 197,252,91,219 // vcvtdq2ps %ymm3,%ymm3
+ .byte 196,98,125,24,74,116 // vbroadcastss 0x74(%rdx),%ymm9
+ .byte 197,52,89,203 // vmulps %ymm3,%ymm9,%ymm9
+ .byte 197,249,110,90,108 // vmovd 0x6c(%rdx),%xmm3
+ .byte 196,227,121,4,219,0 // vpermilps $0x0,%xmm3,%xmm3
+ .byte 196,227,101,24,219,1 // vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ .byte 196,193,100,84,216 // vandps %ymm8,%ymm3,%ymm3
+ .byte 197,252,91,219 // vcvtdq2ps %ymm3,%ymm3
+ .byte 196,98,125,24,82,120 // vbroadcastss 0x78(%rdx),%ymm10
+ .byte 197,44,89,211 // vmulps %ymm3,%ymm10,%ymm10
+ .byte 197,249,110,90,112 // vmovd 0x70(%rdx),%xmm3
+ .byte 196,227,121,4,219,0 // vpermilps $0x0,%xmm3,%xmm3
+ .byte 196,227,101,24,219,1 // vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ .byte 196,193,100,84,216 // vandps %ymm8,%ymm3,%ymm3
+ .byte 197,252,91,219 // vcvtdq2ps %ymm3,%ymm3
+ .byte 196,98,125,24,66,124 // vbroadcastss 0x7c(%rdx),%ymm8
+ .byte 197,188,89,219 // vmulps %ymm3,%ymm8,%ymm3
+ .byte 197,252,92,196 // vsubps %ymm4,%ymm0,%ymm0
+ .byte 196,193,124,89,193 // vmulps %ymm9,%ymm0,%ymm0
+ .byte 197,252,88,196 // vaddps %ymm4,%ymm0,%ymm0
+ .byte 197,244,92,205 // vsubps %ymm5,%ymm1,%ymm1
+ .byte 196,193,116,89,202 // vmulps %ymm10,%ymm1,%ymm1
+ .byte 197,244,88,205 // vaddps %ymm5,%ymm1,%ymm1
+ .byte 197,236,92,214 // vsubps %ymm6,%ymm2,%ymm2
+ .byte 197,236,89,211 // vmulps %ymm3,%ymm2,%ymm2
+ .byte 197,236,88,214 // vaddps %ymm6,%ymm2,%ymm2
+ .byte 196,226,125,24,26 // vbroadcastss (%rdx),%ymm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_load_tables_avx
_sk_load_tables_avx:
.byte 65,87 // push %r15
@@ -3214,6 +3484,18 @@
.byte 72,131,196,24 // add $0x18,%rsp
.byte 255,224 // jmpq *%rax
+.globl _sk_scale_1_float_sse41
+_sk_scale_1_float_sse41:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 69,15,198,192,0 // shufps $0x0,%xmm8,%xmm8
+ .byte 65,15,89,192 // mulps %xmm8,%xmm0
+ .byte 65,15,89,200 // mulps %xmm8,%xmm1
+ .byte 65,15,89,208 // mulps %xmm8,%xmm2
+ .byte 65,15,89,216 // mulps %xmm8,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_scale_u8_sse41
_sk_scale_u8_sse41:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -3230,6 +3512,26 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_1_float_sse41
+_sk_lerp_1_float_sse41:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 69,15,198,192,0 // shufps $0x0,%xmm8,%xmm8
+ .byte 15,92,196 // subps %xmm4,%xmm0
+ .byte 65,15,89,192 // mulps %xmm8,%xmm0
+ .byte 15,88,196 // addps %xmm4,%xmm0
+ .byte 15,92,205 // subps %xmm5,%xmm1
+ .byte 65,15,89,200 // mulps %xmm8,%xmm1
+ .byte 15,88,205 // addps %xmm5,%xmm1
+ .byte 15,92,214 // subps %xmm6,%xmm2
+ .byte 65,15,89,208 // mulps %xmm8,%xmm2
+ .byte 15,88,214 // addps %xmm6,%xmm2
+ .byte 15,92,223 // subps %xmm7,%xmm3
+ .byte 65,15,89,216 // mulps %xmm8,%xmm3
+ .byte 15,88,223 // addps %xmm7,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_lerp_u8_sse41
_sk_lerp_u8_sse41:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -3254,6 +3556,46 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_565_sse41
+_sk_lerp_565_sse41:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 72,139,0 // mov (%rax),%rax
+ .byte 102,68,15,56,51,4,120 // pmovzxwd (%rax,%rdi,2),%xmm8
+ .byte 102,15,110,90,104 // movd 0x68(%rdx),%xmm3
+ .byte 102,15,112,219,0 // pshufd $0x0,%xmm3,%xmm3
+ .byte 102,65,15,219,216 // pand %xmm8,%xmm3
+ .byte 68,15,91,203 // cvtdq2ps %xmm3,%xmm9
+ .byte 243,15,16,26 // movss (%rdx),%xmm3
+ .byte 243,68,15,16,82,116 // movss 0x74(%rdx),%xmm10
+ .byte 69,15,198,210,0 // shufps $0x0,%xmm10,%xmm10
+ .byte 69,15,89,209 // mulps %xmm9,%xmm10
+ .byte 102,68,15,110,74,108 // movd 0x6c(%rdx),%xmm9
+ .byte 102,69,15,112,201,0 // pshufd $0x0,%xmm9,%xmm9
+ .byte 102,69,15,219,200 // pand %xmm8,%xmm9
+ .byte 69,15,91,201 // cvtdq2ps %xmm9,%xmm9
+ .byte 243,68,15,16,90,120 // movss 0x78(%rdx),%xmm11
+ .byte 69,15,198,219,0 // shufps $0x0,%xmm11,%xmm11
+ .byte 69,15,89,217 // mulps %xmm9,%xmm11
+ .byte 102,68,15,110,74,112 // movd 0x70(%rdx),%xmm9
+ .byte 102,69,15,112,201,0 // pshufd $0x0,%xmm9,%xmm9
+ .byte 102,69,15,219,200 // pand %xmm8,%xmm9
+ .byte 69,15,91,193 // cvtdq2ps %xmm9,%xmm8
+ .byte 243,68,15,16,74,124 // movss 0x7c(%rdx),%xmm9
+ .byte 69,15,198,201,0 // shufps $0x0,%xmm9,%xmm9
+ .byte 69,15,89,200 // mulps %xmm8,%xmm9
+ .byte 15,92,196 // subps %xmm4,%xmm0
+ .byte 65,15,89,194 // mulps %xmm10,%xmm0
+ .byte 15,88,196 // addps %xmm4,%xmm0
+ .byte 15,92,205 // subps %xmm5,%xmm1
+ .byte 65,15,89,203 // mulps %xmm11,%xmm1
+ .byte 15,88,205 // addps %xmm5,%xmm1
+ .byte 15,92,214 // subps %xmm6,%xmm2
+ .byte 65,15,89,209 // mulps %xmm9,%xmm2
+ .byte 15,88,214 // addps %xmm6,%xmm2
+ .byte 15,198,219,0 // shufps $0x0,%xmm3,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_load_tables_sse41
_sk_load_tables_sse41:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -4013,6 +4355,18 @@
.byte 72,131,196,40 // add $0x28,%rsp
.byte 255,224 // jmpq *%rax
+.globl _sk_scale_1_float_sse2
+_sk_scale_1_float_sse2:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 69,15,198,192,0 // shufps $0x0,%xmm8,%xmm8
+ .byte 65,15,89,192 // mulps %xmm8,%xmm0
+ .byte 65,15,89,200 // mulps %xmm8,%xmm1
+ .byte 65,15,89,208 // mulps %xmm8,%xmm2
+ .byte 65,15,89,216 // mulps %xmm8,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_scale_u8_sse2
_sk_scale_u8_sse2:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -4032,6 +4386,26 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_1_float_sse2
+_sk_lerp_1_float_sse2:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 69,15,198,192,0 // shufps $0x0,%xmm8,%xmm8
+ .byte 15,92,196 // subps %xmm4,%xmm0
+ .byte 65,15,89,192 // mulps %xmm8,%xmm0
+ .byte 15,88,196 // addps %xmm4,%xmm0
+ .byte 15,92,205 // subps %xmm5,%xmm1
+ .byte 65,15,89,200 // mulps %xmm8,%xmm1
+ .byte 15,88,205 // addps %xmm5,%xmm1
+ .byte 15,92,214 // subps %xmm6,%xmm2
+ .byte 65,15,89,208 // mulps %xmm8,%xmm2
+ .byte 15,88,214 // addps %xmm6,%xmm2
+ .byte 15,92,223 // subps %xmm7,%xmm3
+ .byte 65,15,89,216 // mulps %xmm8,%xmm3
+ .byte 15,88,223 // addps %xmm7,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_lerp_u8_sse2
_sk_lerp_u8_sse2:
.byte 72,173 // lods %ds:(%rsi),%rax
@@ -4059,6 +4433,48 @@
.byte 72,173 // lods %ds:(%rsi),%rax
.byte 255,224 // jmpq *%rax
+.globl _sk_lerp_565_sse2
+_sk_lerp_565_sse2:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 72,139,0 // mov (%rax),%rax
+ .byte 243,68,15,126,4,120 // movq (%rax,%rdi,2),%xmm8
+ .byte 102,15,239,219 // pxor %xmm3,%xmm3
+ .byte 102,68,15,97,195 // punpcklwd %xmm3,%xmm8
+ .byte 102,15,110,90,104 // movd 0x68(%rdx),%xmm3
+ .byte 102,15,112,219,0 // pshufd $0x0,%xmm3,%xmm3
+ .byte 102,65,15,219,216 // pand %xmm8,%xmm3
+ .byte 68,15,91,203 // cvtdq2ps %xmm3,%xmm9
+ .byte 243,15,16,26 // movss (%rdx),%xmm3
+ .byte 243,68,15,16,82,116 // movss 0x74(%rdx),%xmm10
+ .byte 69,15,198,210,0 // shufps $0x0,%xmm10,%xmm10
+ .byte 69,15,89,209 // mulps %xmm9,%xmm10
+ .byte 102,68,15,110,74,108 // movd 0x6c(%rdx),%xmm9
+ .byte 102,69,15,112,201,0 // pshufd $0x0,%xmm9,%xmm9
+ .byte 102,69,15,219,200 // pand %xmm8,%xmm9
+ .byte 69,15,91,201 // cvtdq2ps %xmm9,%xmm9
+ .byte 243,68,15,16,90,120 // movss 0x78(%rdx),%xmm11
+ .byte 69,15,198,219,0 // shufps $0x0,%xmm11,%xmm11
+ .byte 69,15,89,217 // mulps %xmm9,%xmm11
+ .byte 102,68,15,110,74,112 // movd 0x70(%rdx),%xmm9
+ .byte 102,69,15,112,201,0 // pshufd $0x0,%xmm9,%xmm9
+ .byte 102,69,15,219,200 // pand %xmm8,%xmm9
+ .byte 69,15,91,193 // cvtdq2ps %xmm9,%xmm8
+ .byte 243,68,15,16,74,124 // movss 0x7c(%rdx),%xmm9
+ .byte 69,15,198,201,0 // shufps $0x0,%xmm9,%xmm9
+ .byte 69,15,89,200 // mulps %xmm8,%xmm9
+ .byte 15,92,196 // subps %xmm4,%xmm0
+ .byte 65,15,89,194 // mulps %xmm10,%xmm0
+ .byte 15,88,196 // addps %xmm4,%xmm0
+ .byte 15,92,205 // subps %xmm5,%xmm1
+ .byte 65,15,89,203 // mulps %xmm11,%xmm1
+ .byte 15,88,205 // addps %xmm5,%xmm1
+ .byte 15,92,214 // subps %xmm6,%xmm2
+ .byte 65,15,89,209 // mulps %xmm9,%xmm2
+ .byte 15,88,214 // addps %xmm6,%xmm2
+ .byte 15,198,219,0 // shufps $0x0,%xmm3,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
.globl _sk_load_tables_sse2
_sk_load_tables_sse2:
.byte 72,173 // lods %ds:(%rsi),%rax
diff --git a/src/jumper/SkJumper_generated_win.S b/src/jumper/SkJumper_generated_win.S
index 141975a..8d80694 100644
--- a/src/jumper/SkJumper_generated_win.S
+++ b/src/jumper/SkJumper_generated_win.S
@@ -339,6 +339,17 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_scale_1_float_hsw
+_sk_scale_1_float_hsw LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 196,98,125,24,0 ; vbroadcastss (%rax),%ymm8
+ DB 197,188,89,192 ; vmulps %ymm0,%ymm8,%ymm0
+ DB 197,188,89,201 ; vmulps %ymm1,%ymm8,%ymm1
+ DB 197,188,89,210 ; vmulps %ymm2,%ymm8,%ymm2
+ DB 197,188,89,219 ; vmulps %ymm3,%ymm8,%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_scale_u8_hsw
_sk_scale_u8_hsw LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -354,6 +365,21 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_1_float_hsw
+_sk_lerp_1_float_hsw LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 196,98,125,24,0 ; vbroadcastss (%rax),%ymm8
+ DB 197,252,92,196 ; vsubps %ymm4,%ymm0,%ymm0
+ DB 196,226,61,168,196 ; vfmadd213ps %ymm4,%ymm8,%ymm0
+ DB 197,244,92,205 ; vsubps %ymm5,%ymm1,%ymm1
+ DB 196,226,61,168,205 ; vfmadd213ps %ymm5,%ymm8,%ymm1
+ DB 197,236,92,214 ; vsubps %ymm6,%ymm2,%ymm2
+ DB 196,226,61,168,214 ; vfmadd213ps %ymm6,%ymm8,%ymm2
+ DB 197,228,92,223 ; vsubps %ymm7,%ymm3,%ymm3
+ DB 196,226,61,168,223 ; vfmadd213ps %ymm7,%ymm8,%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_lerp_u8_hsw
_sk_lerp_u8_hsw LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -373,6 +399,36 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_565_hsw
+_sk_lerp_565_hsw LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 72,139,0 ; mov (%rax),%rax
+ DB 196,226,125,51,28,120 ; vpmovzxwd (%rax,%rdi,2),%ymm3
+ DB 196,98,125,88,66,104 ; vpbroadcastd 0x68(%rdx),%ymm8
+ DB 197,61,219,195 ; vpand %ymm3,%ymm8,%ymm8
+ DB 196,65,124,91,192 ; vcvtdq2ps %ymm8,%ymm8
+ DB 196,98,125,24,74,116 ; vbroadcastss 0x74(%rdx),%ymm9
+ DB 196,65,52,89,192 ; vmulps %ymm8,%ymm9,%ymm8
+ DB 196,98,125,88,74,108 ; vpbroadcastd 0x6c(%rdx),%ymm9
+ DB 197,53,219,203 ; vpand %ymm3,%ymm9,%ymm9
+ DB 196,65,124,91,201 ; vcvtdq2ps %ymm9,%ymm9
+ DB 196,98,125,24,82,120 ; vbroadcastss 0x78(%rdx),%ymm10
+ DB 196,65,44,89,201 ; vmulps %ymm9,%ymm10,%ymm9
+ DB 196,98,125,88,82,112 ; vpbroadcastd 0x70(%rdx),%ymm10
+ DB 197,173,219,219 ; vpand %ymm3,%ymm10,%ymm3
+ DB 197,252,91,219 ; vcvtdq2ps %ymm3,%ymm3
+ DB 196,98,125,24,82,124 ; vbroadcastss 0x7c(%rdx),%ymm10
+ DB 197,172,89,219 ; vmulps %ymm3,%ymm10,%ymm3
+ DB 197,252,92,196 ; vsubps %ymm4,%ymm0,%ymm0
+ DB 196,226,61,168,196 ; vfmadd213ps %ymm4,%ymm8,%ymm0
+ DB 197,244,92,205 ; vsubps %ymm5,%ymm1,%ymm1
+ DB 196,226,53,168,205 ; vfmadd213ps %ymm5,%ymm9,%ymm1
+ DB 197,236,92,214 ; vsubps %ymm6,%ymm2,%ymm2
+ DB 196,226,101,168,214 ; vfmadd213ps %ymm6,%ymm3,%ymm2
+ DB 196,226,125,24,26 ; vbroadcastss (%rdx),%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_load_tables_hsw
_sk_load_tables_hsw LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -987,6 +1043,17 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_scale_1_float_avx
+_sk_scale_1_float_avx LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 196,98,125,24,0 ; vbroadcastss (%rax),%ymm8
+ DB 197,188,89,192 ; vmulps %ymm0,%ymm8,%ymm0
+ DB 197,188,89,201 ; vmulps %ymm1,%ymm8,%ymm1
+ DB 197,188,89,210 ; vmulps %ymm2,%ymm8,%ymm2
+ DB 197,188,89,219 ; vmulps %ymm3,%ymm8,%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_scale_u8_avx
_sk_scale_u8_avx LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -1004,6 +1071,25 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_1_float_avx
+_sk_lerp_1_float_avx LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 196,98,125,24,0 ; vbroadcastss (%rax),%ymm8
+ DB 197,252,92,196 ; vsubps %ymm4,%ymm0,%ymm0
+ DB 196,193,124,89,192 ; vmulps %ymm8,%ymm0,%ymm0
+ DB 197,252,88,196 ; vaddps %ymm4,%ymm0,%ymm0
+ DB 197,244,92,205 ; vsubps %ymm5,%ymm1,%ymm1
+ DB 196,193,116,89,200 ; vmulps %ymm8,%ymm1,%ymm1
+ DB 197,244,88,205 ; vaddps %ymm5,%ymm1,%ymm1
+ DB 197,236,92,214 ; vsubps %ymm6,%ymm2,%ymm2
+ DB 196,193,108,89,208 ; vmulps %ymm8,%ymm2,%ymm2
+ DB 197,236,88,214 ; vaddps %ymm6,%ymm2,%ymm2
+ DB 197,228,92,223 ; vsubps %ymm7,%ymm3,%ymm3
+ DB 196,193,100,89,216 ; vmulps %ymm8,%ymm3,%ymm3
+ DB 197,228,88,223 ; vaddps %ymm7,%ymm3,%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_lerp_u8_avx
_sk_lerp_u8_avx LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -1029,6 +1115,47 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_565_avx
+_sk_lerp_565_avx LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 72,139,0 ; mov (%rax),%rax
+ DB 196,226,121,51,92,120,8 ; vpmovzxwd 0x8(%rax,%rdi,2),%xmm3
+ DB 196,98,121,51,4,120 ; vpmovzxwd (%rax,%rdi,2),%xmm8
+ DB 196,99,61,24,195,1 ; vinsertf128 $0x1,%xmm3,%ymm8,%ymm8
+ DB 197,249,110,90,104 ; vmovd 0x68(%rdx),%xmm3
+ DB 196,227,121,4,219,0 ; vpermilps $0x0,%xmm3,%xmm3
+ DB 196,227,101,24,219,1 ; vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ DB 196,193,100,84,216 ; vandps %ymm8,%ymm3,%ymm3
+ DB 197,252,91,219 ; vcvtdq2ps %ymm3,%ymm3
+ DB 196,98,125,24,74,116 ; vbroadcastss 0x74(%rdx),%ymm9
+ DB 197,52,89,203 ; vmulps %ymm3,%ymm9,%ymm9
+ DB 197,249,110,90,108 ; vmovd 0x6c(%rdx),%xmm3
+ DB 196,227,121,4,219,0 ; vpermilps $0x0,%xmm3,%xmm3
+ DB 196,227,101,24,219,1 ; vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ DB 196,193,100,84,216 ; vandps %ymm8,%ymm3,%ymm3
+ DB 197,252,91,219 ; vcvtdq2ps %ymm3,%ymm3
+ DB 196,98,125,24,82,120 ; vbroadcastss 0x78(%rdx),%ymm10
+ DB 197,44,89,211 ; vmulps %ymm3,%ymm10,%ymm10
+ DB 197,249,110,90,112 ; vmovd 0x70(%rdx),%xmm3
+ DB 196,227,121,4,219,0 ; vpermilps $0x0,%xmm3,%xmm3
+ DB 196,227,101,24,219,1 ; vinsertf128 $0x1,%xmm3,%ymm3,%ymm3
+ DB 196,193,100,84,216 ; vandps %ymm8,%ymm3,%ymm3
+ DB 197,252,91,219 ; vcvtdq2ps %ymm3,%ymm3
+ DB 196,98,125,24,66,124 ; vbroadcastss 0x7c(%rdx),%ymm8
+ DB 197,188,89,219 ; vmulps %ymm3,%ymm8,%ymm3
+ DB 197,252,92,196 ; vsubps %ymm4,%ymm0,%ymm0
+ DB 196,193,124,89,193 ; vmulps %ymm9,%ymm0,%ymm0
+ DB 197,252,88,196 ; vaddps %ymm4,%ymm0,%ymm0
+ DB 197,244,92,205 ; vsubps %ymm5,%ymm1,%ymm1
+ DB 196,193,116,89,202 ; vmulps %ymm10,%ymm1,%ymm1
+ DB 197,244,88,205 ; vaddps %ymm5,%ymm1,%ymm1
+ DB 197,236,92,214 ; vsubps %ymm6,%ymm2,%ymm2
+ DB 197,236,89,211 ; vmulps %ymm3,%ymm2,%ymm2
+ DB 197,236,88,214 ; vaddps %ymm6,%ymm2,%ymm2
+ DB 196,226,125,24,26 ; vbroadcastss (%rdx),%ymm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_load_tables_avx
_sk_load_tables_avx LABEL PROC
DB 65,87 ; push %r15
@@ -1892,6 +2019,18 @@
DB 72,131,196,24 ; add $0x18,%rsp
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_scale_1_float_sse41
+_sk_scale_1_float_sse41 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 69,15,198,192,0 ; shufps $0x0,%xmm8,%xmm8
+ DB 65,15,89,192 ; mulps %xmm8,%xmm0
+ DB 65,15,89,200 ; mulps %xmm8,%xmm1
+ DB 65,15,89,208 ; mulps %xmm8,%xmm2
+ DB 65,15,89,216 ; mulps %xmm8,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_scale_u8_sse41
_sk_scale_u8_sse41 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -1908,6 +2047,26 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_1_float_sse41
+_sk_lerp_1_float_sse41 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 69,15,198,192,0 ; shufps $0x0,%xmm8,%xmm8
+ DB 15,92,196 ; subps %xmm4,%xmm0
+ DB 65,15,89,192 ; mulps %xmm8,%xmm0
+ DB 15,88,196 ; addps %xmm4,%xmm0
+ DB 15,92,205 ; subps %xmm5,%xmm1
+ DB 65,15,89,200 ; mulps %xmm8,%xmm1
+ DB 15,88,205 ; addps %xmm5,%xmm1
+ DB 15,92,214 ; subps %xmm6,%xmm2
+ DB 65,15,89,208 ; mulps %xmm8,%xmm2
+ DB 15,88,214 ; addps %xmm6,%xmm2
+ DB 15,92,223 ; subps %xmm7,%xmm3
+ DB 65,15,89,216 ; mulps %xmm8,%xmm3
+ DB 15,88,223 ; addps %xmm7,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_lerp_u8_sse41
_sk_lerp_u8_sse41 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -1932,6 +2091,46 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_565_sse41
+_sk_lerp_565_sse41 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 72,139,0 ; mov (%rax),%rax
+ DB 102,68,15,56,51,4,120 ; pmovzxwd (%rax,%rdi,2),%xmm8
+ DB 102,15,110,90,104 ; movd 0x68(%rdx),%xmm3
+ DB 102,15,112,219,0 ; pshufd $0x0,%xmm3,%xmm3
+ DB 102,65,15,219,216 ; pand %xmm8,%xmm3
+ DB 68,15,91,203 ; cvtdq2ps %xmm3,%xmm9
+ DB 243,15,16,26 ; movss (%rdx),%xmm3
+ DB 243,68,15,16,82,116 ; movss 0x74(%rdx),%xmm10
+ DB 69,15,198,210,0 ; shufps $0x0,%xmm10,%xmm10
+ DB 69,15,89,209 ; mulps %xmm9,%xmm10
+ DB 102,68,15,110,74,108 ; movd 0x6c(%rdx),%xmm9
+ DB 102,69,15,112,201,0 ; pshufd $0x0,%xmm9,%xmm9
+ DB 102,69,15,219,200 ; pand %xmm8,%xmm9
+ DB 69,15,91,201 ; cvtdq2ps %xmm9,%xmm9
+ DB 243,68,15,16,90,120 ; movss 0x78(%rdx),%xmm11
+ DB 69,15,198,219,0 ; shufps $0x0,%xmm11,%xmm11
+ DB 69,15,89,217 ; mulps %xmm9,%xmm11
+ DB 102,68,15,110,74,112 ; movd 0x70(%rdx),%xmm9
+ DB 102,69,15,112,201,0 ; pshufd $0x0,%xmm9,%xmm9
+ DB 102,69,15,219,200 ; pand %xmm8,%xmm9
+ DB 69,15,91,193 ; cvtdq2ps %xmm9,%xmm8
+ DB 243,68,15,16,74,124 ; movss 0x7c(%rdx),%xmm9
+ DB 69,15,198,201,0 ; shufps $0x0,%xmm9,%xmm9
+ DB 69,15,89,200 ; mulps %xmm8,%xmm9
+ DB 15,92,196 ; subps %xmm4,%xmm0
+ DB 65,15,89,194 ; mulps %xmm10,%xmm0
+ DB 15,88,196 ; addps %xmm4,%xmm0
+ DB 15,92,205 ; subps %xmm5,%xmm1
+ DB 65,15,89,203 ; mulps %xmm11,%xmm1
+ DB 15,88,205 ; addps %xmm5,%xmm1
+ DB 15,92,214 ; subps %xmm6,%xmm2
+ DB 65,15,89,209 ; mulps %xmm9,%xmm2
+ DB 15,88,214 ; addps %xmm6,%xmm2
+ DB 15,198,219,0 ; shufps $0x0,%xmm3,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_load_tables_sse41
_sk_load_tables_sse41 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -2718,6 +2917,18 @@
DB 72,131,196,40 ; add $0x28,%rsp
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_scale_1_float_sse2
+_sk_scale_1_float_sse2 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 69,15,198,192,0 ; shufps $0x0,%xmm8,%xmm8
+ DB 65,15,89,192 ; mulps %xmm8,%xmm0
+ DB 65,15,89,200 ; mulps %xmm8,%xmm1
+ DB 65,15,89,208 ; mulps %xmm8,%xmm2
+ DB 65,15,89,216 ; mulps %xmm8,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_scale_u8_sse2
_sk_scale_u8_sse2 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -2737,6 +2948,26 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_1_float_sse2
+_sk_lerp_1_float_sse2 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 69,15,198,192,0 ; shufps $0x0,%xmm8,%xmm8
+ DB 15,92,196 ; subps %xmm4,%xmm0
+ DB 65,15,89,192 ; mulps %xmm8,%xmm0
+ DB 15,88,196 ; addps %xmm4,%xmm0
+ DB 15,92,205 ; subps %xmm5,%xmm1
+ DB 65,15,89,200 ; mulps %xmm8,%xmm1
+ DB 15,88,205 ; addps %xmm5,%xmm1
+ DB 15,92,214 ; subps %xmm6,%xmm2
+ DB 65,15,89,208 ; mulps %xmm8,%xmm2
+ DB 15,88,214 ; addps %xmm6,%xmm2
+ DB 15,92,223 ; subps %xmm7,%xmm3
+ DB 65,15,89,216 ; mulps %xmm8,%xmm3
+ DB 15,88,223 ; addps %xmm7,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_lerp_u8_sse2
_sk_lerp_u8_sse2 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
@@ -2764,6 +2995,48 @@
DB 72,173 ; lods %ds:(%rsi),%rax
DB 255,224 ; jmpq *%rax
+PUBLIC _sk_lerp_565_sse2
+_sk_lerp_565_sse2 LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 72,139,0 ; mov (%rax),%rax
+ DB 243,68,15,126,4,120 ; movq (%rax,%rdi,2),%xmm8
+ DB 102,15,239,219 ; pxor %xmm3,%xmm3
+ DB 102,68,15,97,195 ; punpcklwd %xmm3,%xmm8
+ DB 102,15,110,90,104 ; movd 0x68(%rdx),%xmm3
+ DB 102,15,112,219,0 ; pshufd $0x0,%xmm3,%xmm3
+ DB 102,65,15,219,216 ; pand %xmm8,%xmm3
+ DB 68,15,91,203 ; cvtdq2ps %xmm3,%xmm9
+ DB 243,15,16,26 ; movss (%rdx),%xmm3
+ DB 243,68,15,16,82,116 ; movss 0x74(%rdx),%xmm10
+ DB 69,15,198,210,0 ; shufps $0x0,%xmm10,%xmm10
+ DB 69,15,89,209 ; mulps %xmm9,%xmm10
+ DB 102,68,15,110,74,108 ; movd 0x6c(%rdx),%xmm9
+ DB 102,69,15,112,201,0 ; pshufd $0x0,%xmm9,%xmm9
+ DB 102,69,15,219,200 ; pand %xmm8,%xmm9
+ DB 69,15,91,201 ; cvtdq2ps %xmm9,%xmm9
+ DB 243,68,15,16,90,120 ; movss 0x78(%rdx),%xmm11
+ DB 69,15,198,219,0 ; shufps $0x0,%xmm11,%xmm11
+ DB 69,15,89,217 ; mulps %xmm9,%xmm11
+ DB 102,68,15,110,74,112 ; movd 0x70(%rdx),%xmm9
+ DB 102,69,15,112,201,0 ; pshufd $0x0,%xmm9,%xmm9
+ DB 102,69,15,219,200 ; pand %xmm8,%xmm9
+ DB 69,15,91,193 ; cvtdq2ps %xmm9,%xmm8
+ DB 243,68,15,16,74,124 ; movss 0x7c(%rdx),%xmm9
+ DB 69,15,198,201,0 ; shufps $0x0,%xmm9,%xmm9
+ DB 69,15,89,200 ; mulps %xmm8,%xmm9
+ DB 15,92,196 ; subps %xmm4,%xmm0
+ DB 65,15,89,194 ; mulps %xmm10,%xmm0
+ DB 15,88,196 ; addps %xmm4,%xmm0
+ DB 15,92,205 ; subps %xmm5,%xmm1
+ DB 65,15,89,203 ; mulps %xmm11,%xmm1
+ DB 15,88,205 ; addps %xmm5,%xmm1
+ DB 15,92,214 ; subps %xmm6,%xmm2
+ DB 65,15,89,209 ; mulps %xmm9,%xmm2
+ DB 15,88,214 ; addps %xmm6,%xmm2
+ DB 15,198,219,0 ; shufps $0x0,%xmm3,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
PUBLIC _sk_load_tables_sse2
_sk_load_tables_sse2 LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
diff --git a/src/jumper/SkJumper_stages.cpp b/src/jumper/SkJumper_stages.cpp
index 6caf058..580432c 100644
--- a/src/jumper/SkJumper_stages.cpp
+++ b/src/jumper/SkJumper_stages.cpp
@@ -460,6 +460,14 @@
b = fn(b);
}
+STAGE(scale_1_float) {
+ auto c = *(const float*)ctx;
+
+ r = r * c;
+ g = g * c;
+ b = b * c;
+ a = a * c;
+}
STAGE(scale_u8) {
auto ptr = *(const uint8_t**)ctx + x;
@@ -471,6 +479,15 @@
b = b * c;
a = a * c;
}
+
+STAGE(lerp_1_float) {
+ auto c = *(const float*)ctx;
+
+ r = lerp(dr, r, c);
+ g = lerp(dg, g, c);
+ b = lerp(db, b, c);
+ a = lerp(da, a, c);
+}
STAGE(lerp_u8) {
auto ptr = *(const uint8_t**)ctx + x;
@@ -482,6 +499,17 @@
b = lerp(db, b, c);
a = lerp(da, a, c);
}
+STAGE(lerp_565) {
+ auto ptr = *(const uint16_t**)ctx + x;
+
+ F cr,cg,cb;
+ from_565(unaligned_load<U16>(ptr), &cr, &cg, &cb, k);
+
+ r = lerp(dr, r, cr);
+ g = lerp(dg, g, cg);
+ b = lerp(db, b, cb);
+ a = k->_1;
+}
STAGE(load_tables) {
struct Ctx {
@@ -500,8 +528,7 @@
STAGE(load_565) {
auto ptr = *(const uint16_t**)ctx + x;
- auto px = unaligned_load<U16>(ptr);
- from_565(px, &r,&g,&b, k);
+ from_565(unaligned_load<U16>(ptr), &r,&g,&b, k);
a = k->_1;
}
STAGE(store_565) {
