SkColorCubeFilter: require alpha == 0xFF.

This is about a 12% improvement on my desktop, from 134 to 118ms on our bench.

BUG=skia:

Review URL: https://codereview.chromium.org/1295873004

5 files changed, 13 insertions, 13 deletions

diff --git a/src/core/SkPMFloat.h b/src/core/SkPMFloat.h
index 1fc80f5004..4a5621f5b3 100644
--- a/src/core/SkPMFloat.h
+++ b/src/core/SkPMFloat.h
@@ -26,7 +26,7 @@ class SkPMFloat : public Sk4f {
 public:
     static SkPMFloat FromPMColor(SkPMColor c) { return SkPMFloat(c); }
     static SkPMFloat FromARGB(float a, float r, float g, float b) { return SkPMFloat(a,r,g,b); }
-    static SkPMFloat FromBGRx(SkColor c);  // Ignores c's alpha, instead forcing it to 1.
+    static SkPMFloat FromOpaqueColor(SkColor c);  // Requires c's alpha == 0xFF.
 
     Sk4f alphas() const;  // argb -> aaaa, generally faster than the equivalent Sk4f(this->a()).
 
diff --git a/src/opts/SkColorCubeFilter_opts.h b/src/opts/SkColorCubeFilter_opts.h
index 92c75664d2..a61f66d174 100644
--- a/src/opts/SkColorCubeFilter_opts.h
+++ b/src/opts/SkColorCubeFilter_opts.h
@@ -61,10 +61,10 @@ void color_cube_filter_span(const SkPMColor src[],
             const SkColor lutColor10 = colorCube[ix + i10];
             const SkColor lutColor11 = colorCube[ix + i11];
 
-            Sk4f  sum = SkPMFloat::FromBGRx(lutColor00) * g0b0;
-            sum = sum + SkPMFloat::FromBGRx(lutColor01) * g0b1;
-            sum = sum + SkPMFloat::FromBGRx(lutColor10) * g1b0;
-            sum = sum + SkPMFloat::FromBGRx(lutColor11) * g1b1;
+            Sk4f  sum = SkPMFloat::FromOpaqueColor(lutColor00) * g0b0;
+            sum = sum + SkPMFloat::FromOpaqueColor(lutColor01) * g0b1;
+            sum = sum + SkPMFloat::FromOpaqueColor(lutColor10) * g1b0;
+            sum = sum + SkPMFloat::FromOpaqueColor(lutColor11) * g1b1;
 
             color = color + sum * Sk4f((float)colorToFactors[x][r]);
         }
diff --git a/src/opts/SkPMFloat_neon.h b/src/opts/SkPMFloat_neon.h
index ecb151f1fd..4e099f91ec 100644
--- a/src/opts/SkPMFloat_neon.h
+++ b/src/opts/SkPMFloat_neon.h
@@ -33,14 +33,14 @@ inline Sk4f SkPMFloat::alphas() const {
     return vdupq_lane_f32(vget_high_f32(fVec), 1);  // Duplicate high lane of high half i.e. lane 3.
 }
 
-inline SkPMFloat SkPMFloat::FromBGRx(SkColor c) {
+inline SkPMFloat SkPMFloat::FromOpaqueColor(SkColor c) {
+    SkASSERT(SkColorGetA(c) == 0xFF);
     uint8x8_t fix8 = (uint8x8_t)vdup_n_u32(c);
 #if defined(SK_PMCOLOR_IS_RGBA)
     fix8 = vtbl1_u8(fix8, vcreate_u8(0x0300010203000102ULL));  // 03 00 01 02, 2x, i.e. swap R&B.
 #endif
     uint16x8_t fix8_16 = vmovl_u8(fix8);
     uint32x4_t fix8_32 = vmovl_u16(vget_low_u16(fix8_16));
-    fix8_32 = vsetq_lane_u32(0xFF, fix8_32, 3);  // Force alpha to 1.
 
     SkPMFloat pmf = Sk4f(vmulq_f32(vcvtq_f32_u32(fix8_32), vdupq_n_f32(1.0f/255)));
     SkASSERT(pmf.isValid());
diff --git a/src/opts/SkPMFloat_none.h b/src/opts/SkPMFloat_none.h
index adc0bde3b0..17c76d90f4 100644
--- a/src/opts/SkPMFloat_none.h
+++ b/src/opts/SkPMFloat_none.h
@@ -34,7 +34,8 @@ inline Sk4f SkPMFloat::alphas() const {
     return Sk4f(this->a());
 }
 
-inline SkPMFloat SkPMFloat::FromBGRx(SkColor c) {
+inline SkPMFloat SkPMFloat::FromOpaqueColor(SkColor c) {
+  SkASSERT(SkColorGetA(c) == 0xFF);
   float inv255 = 1.0f / 255;
   SkPMFloat pmf = SkPMFloat::FromARGB(1.0f,
                                       SkColorGetR(c) * inv255,
diff --git a/src/opts/SkPMFloat_sse.h b/src/opts/SkPMFloat_sse.h
index 6cfee1da6f..6a4d5b6b5c 100644
--- a/src/opts/SkPMFloat_sse.h
+++ b/src/opts/SkPMFloat_sse.h
@@ -38,15 +38,16 @@ inline Sk4f SkPMFloat::alphas() const {
     return _mm_shuffle_ps(fVec, fVec, 0xff);  // Read as 11 11 11 11, copying lane 3 to all lanes.
 }
 
-inline SkPMFloat SkPMFloat::FromBGRx(SkColor c) {
+inline SkPMFloat SkPMFloat::FromOpaqueColor(SkColor c) {
+    SkASSERT(SkColorGetA(c) == 0xFF);
     __m128i fix8 = _mm_cvtsi32_si128((int)c);
 #if SK_CPU_SSE_LEVEL >= SK_CPU_SSE_LEVEL_SSSE3
     const char _ = ~0;  // Zero these bytes.
     __m128i fix8_32 = _mm_shuffle_epi8(fix8,
     #if defined(SK_PMCOLOR_IS_BGRA)
-            _mm_setr_epi8(0,_,_,_, 1,_,_,_, 2,_,_,_, _,_,_,_)
+            _mm_setr_epi8(0,_,_,_, 1,_,_,_, 2,_,_,_, 3,_,_,_)
     #else
-            _mm_setr_epi8(2,_,_,_, 1,_,_,_, 0,_,_,_, _,_,_,_)
+            _mm_setr_epi8(2,_,_,_, 1,_,_,_, 0,_,_,_, 3,_,_,_)
     #endif
     );
 #else
@@ -56,8 +57,6 @@ inline SkPMFloat SkPMFloat::FromBGRx(SkColor c) {
         fix8_32 = _mm_shuffle_epi32(fix8_32, 0xC6);  // C6 == 11 00 01 10, i.e swap lanes 0 and 2.
     #endif
 #endif
-    fix8_32 = _mm_or_si128(fix8_32, _mm_set_epi32(0xFF,0,0,0));  // Force alpha to 1.
-
     SkPMFloat pmf = Sk4f(_mm_mul_ps(_mm_cvtepi32_ps(fix8_32), _mm_set1_ps(1.0f/255)));
     SkASSERT(pmf.isValid());
     return pmf;