aboutsummaryrefslogtreecommitdiffstats
path: root/src/opts
diff options
context:
space:
mode:
Diffstat (limited to 'src/opts')
-rw-r--r--src/opts/SkNx_neon.h120
-rw-r--r--src/opts/SkNx_sse.h94
-rw-r--r--src/opts/SkXfermode_opts_SSE2.cpp8
-rw-r--r--src/opts/SkXfermode_opts_arm_neon.cpp1
4 files changed, 103 insertions, 120 deletions
diff --git a/src/opts/SkNx_neon.h b/src/opts/SkNx_neon.h
index da926e0b4c..b319807779 100644
--- a/src/opts/SkNx_neon.h
+++ b/src/opts/SkNx_neon.h
@@ -33,34 +33,7 @@ namespace { // See SkNx.h
case 31: return op(v, 31); } return fVec
template <>
-class SkNb<2, 4> {
-public:
- SkNb(uint32x2_t vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return vget_lane_u32(fVec, 0) && vget_lane_u32(fVec, 1); }
- bool anyTrue() const { return vget_lane_u32(fVec, 0) || vget_lane_u32(fVec, 1); }
-
- uint32x2_t fVec;
-};
-
-template <>
-class SkNb<4, 4> {
-public:
- SkNb(uint32x4_t vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return vgetq_lane_u32(fVec, 0) && vgetq_lane_u32(fVec, 1)
- && vgetq_lane_u32(fVec, 2) && vgetq_lane_u32(fVec, 3); }
- bool anyTrue() const { return vgetq_lane_u32(fVec, 0) || vgetq_lane_u32(fVec, 1)
- || vgetq_lane_u32(fVec, 2) || vgetq_lane_u32(fVec, 3); }
-
- uint32x4_t fVec;
-};
-
-template <>
class SkNf<2, float> {
- typedef SkNb<2, 4> Nb;
public:
SkNf(float32x2_t vec) : fVec(vec) {}
@@ -93,12 +66,14 @@ public:
#endif
}
- Nb operator == (const SkNf& o) const { return vceq_f32(fVec, o.fVec); }
- Nb operator < (const SkNf& o) const { return vclt_f32(fVec, o.fVec); }
- Nb operator > (const SkNf& o) const { return vcgt_f32(fVec, o.fVec); }
- Nb operator <= (const SkNf& o) const { return vcle_f32(fVec, o.fVec); }
- Nb operator >= (const SkNf& o) const { return vcge_f32(fVec, o.fVec); }
- Nb operator != (const SkNf& o) const { return vmvn_u32(vceq_f32(fVec, o.fVec)); }
+ SkNf operator == (const SkNf& o) const { return vreinterpret_f32_u32(vceq_f32(fVec, o.fVec)); }
+ SkNf operator < (const SkNf& o) const { return vreinterpret_f32_u32(vclt_f32(fVec, o.fVec)); }
+ SkNf operator > (const SkNf& o) const { return vreinterpret_f32_u32(vcgt_f32(fVec, o.fVec)); }
+ SkNf operator <= (const SkNf& o) const { return vreinterpret_f32_u32(vcle_f32(fVec, o.fVec)); }
+ SkNf operator >= (const SkNf& o) const { return vreinterpret_f32_u32(vcge_f32(fVec, o.fVec)); }
+ SkNf operator != (const SkNf& o) const {
+ return vreinterpret_f32_u32(vmvn_u32(vceq_f32(fVec, o.fVec)));
+ }
static SkNf Min(const SkNf& l, const SkNf& r) { return vmin_f32(l.fVec, r.fVec); }
static SkNf Max(const SkNf& l, const SkNf& r) { return vmax_f32(l.fVec, r.fVec); }
@@ -126,25 +101,21 @@ public:
return vget_lane_f32(fVec, k&1);
}
+ bool allTrue() const {
+ auto v = vreinterpret_u32_f32(fVec);
+ return vget_lane_u32(v,0) && vget_lane_u32(v,1);
+ }
+ bool anyTrue() const {
+ auto v = vreinterpret_u32_f32(fVec);
+ return vget_lane_u32(v,0) || vget_lane_u32(v,1);
+ }
+
float32x2_t fVec;
};
#if defined(SK_CPU_ARM64)
template <>
-class SkNb<2, 8> {
-public:
- SkNb(uint64x2_t vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return vgetq_lane_u64(fVec, 0) && vgetq_lane_u64(fVec, 1); }
- bool anyTrue() const { return vgetq_lane_u64(fVec, 0) || vgetq_lane_u64(fVec, 1); }
-
- uint64x2_t fVec;
-};
-
-template <>
class SkNf<2, double> {
- typedef SkNb<2, 8> Nb;
public:
SkNf(float64x2_t vec) : fVec(vec) {}
@@ -160,13 +131,14 @@ public:
SkNf operator * (const SkNf& o) const { return vmulq_f64(fVec, o.fVec); }
SkNf operator / (const SkNf& o) const { return vdivq_f64(fVec, o.fVec); }
- Nb operator == (const SkNf& o) const { return vceqq_f64(fVec, o.fVec); }
- Nb operator < (const SkNf& o) const { return vcltq_f64(fVec, o.fVec); }
- Nb operator > (const SkNf& o) const { return vcgtq_f64(fVec, o.fVec); }
- Nb operator <= (const SkNf& o) const { return vcleq_f64(fVec, o.fVec); }
- Nb operator >= (const SkNf& o) const { return vcgeq_f64(fVec, o.fVec); }
- Nb operator != (const SkNf& o) const {
- return vreinterpretq_u64_u32(vmvnq_u32(vreinterpretq_u32_u64(vceqq_f64(fVec, o.fVec))));
+ // vreinterpretq_f64_u64 and vreinterpretq_f64_u32 don't seem to exist.... weird.
+ SkNf operator==(const SkNf& o) const { return (float64x2_t)(vceqq_f64(fVec, o.fVec)); }
+ SkNf operator <(const SkNf& o) const { return (float64x2_t)(vcltq_f64(fVec, o.fVec)); }
+ SkNf operator >(const SkNf& o) const { return (float64x2_t)(vcgtq_f64(fVec, o.fVec)); }
+ SkNf operator<=(const SkNf& o) const { return (float64x2_t)(vcleq_f64(fVec, o.fVec)); }
+ SkNf operator>=(const SkNf& o) const { return (float64x2_t)(vcgeq_f64(fVec, o.fVec)); }
+ SkNf operator != (const SkNf& o) const {
+ return (float64x2_t)(vmvnq_u32(vreinterpretq_u32_u64(vceqq_f64(fVec, o.fVec))));
}
static SkNf Min(const SkNf& l, const SkNf& r) { return vminq_f64(l.fVec, r.fVec); }
@@ -202,6 +174,16 @@ public:
return vgetq_lane_f64(fVec, k&1);
}
+ // vreinterpretq_u64_f64 doesn't seem to exist.... weird.
+ bool allTrue() const {
+ auto v = (uint64x2_t)(fVec);
+ return vgetq_lane_u64(v,0) && vgetq_lane_u64(v,1);
+ }
+ bool anyTrue() const {
+ auto v = (uint64x2_t)(fVec);
+ return vgetq_lane_u64(v,0) || vgetq_lane_u64(v,1);
+ }
+
float64x2_t fVec;
};
#endif//defined(SK_CPU_ARM64)
@@ -235,7 +217,6 @@ public:
template <>
class SkNf<4, float> {
- typedef SkNb<4, 4> Nb;
public:
SkNf(float32x4_t vec) : fVec(vec) {}
@@ -270,12 +251,14 @@ public:
#endif
}
- Nb operator == (const SkNf& o) const { return vceqq_f32(fVec, o.fVec); }
- Nb operator < (const SkNf& o) const { return vcltq_f32(fVec, o.fVec); }
- Nb operator > (const SkNf& o) const { return vcgtq_f32(fVec, o.fVec); }
- Nb operator <= (const SkNf& o) const { return vcleq_f32(fVec, o.fVec); }
- Nb operator >= (const SkNf& o) const { return vcgeq_f32(fVec, o.fVec); }
- Nb operator != (const SkNf& o) const { return vmvnq_u32(vceqq_f32(fVec, o.fVec)); }
+ SkNf operator==(const SkNf& o) const { return vreinterpretq_f32_u32(vceqq_f32(fVec, o.fVec)); }
+ SkNf operator <(const SkNf& o) const { return vreinterpretq_f32_u32(vcltq_f32(fVec, o.fVec)); }
+ SkNf operator >(const SkNf& o) const { return vreinterpretq_f32_u32(vcgtq_f32(fVec, o.fVec)); }
+ SkNf operator<=(const SkNf& o) const { return vreinterpretq_f32_u32(vcleq_f32(fVec, o.fVec)); }
+ SkNf operator>=(const SkNf& o) const { return vreinterpretq_f32_u32(vcgeq_f32(fVec, o.fVec)); }
+ SkNf operator!=(const SkNf& o) const {
+ return vreinterpretq_f32_u32(vmvnq_u32(vceqq_f32(fVec, o.fVec)));
+ }
static SkNf Min(const SkNf& l, const SkNf& r) { return vminq_f32(l.fVec, r.fVec); }
static SkNf Max(const SkNf& l, const SkNf& r) { return vmaxq_f32(l.fVec, r.fVec); }
@@ -303,6 +286,17 @@ public:
return vgetq_lane_f32(fVec, k&3);
}
+ bool allTrue() const {
+ auto v = vreinterpretq_u32_f32(fVec);
+ return vgetq_lane_u32(v,0) && vgetq_lane_u32(v,1)
+ && vgetq_lane_u32(v,2) && vgetq_lane_u32(v,3);
+ }
+ bool anyTrue() const {
+ auto v = vreinterpretq_u32_f32(fVec);
+ return vgetq_lane_u32(v,0) || vgetq_lane_u32(v,1)
+ || vgetq_lane_u32(v,2) || vgetq_lane_u32(v,3);
+ }
+
float32x4_t fVec;
};
@@ -363,12 +357,18 @@ public:
SkNi operator - (const SkNi& o) const { return vsubq_u8(fVec, o.fVec); }
static SkNi Min(const SkNi& a, const SkNi& b) { return vminq_u8(a.fVec, b.fVec); }
+ SkNi operator < (const SkNi& o) const { return vcltq_u8(fVec, o.fVec); }
template <int k> uint8_t kth() const {
SkASSERT(0 <= k && k < 15);
return vgetq_lane_u8(fVec, k&16);
}
+ SkNi thenElse(const SkNi& t, const SkNi& e) const {
+ return vorrq_u8(vandq_u8(t.fVec, fVec),
+ vbicq_u8(e.fVec, fVec));
+ }
+
uint8x16_t fVec;
};
diff --git a/src/opts/SkNx_sse.h b/src/opts/SkNx_sse.h
index 12a4719d0c..9b4de700ee 100644
--- a/src/opts/SkNx_sse.h
+++ b/src/opts/SkNx_sse.h
@@ -12,46 +12,9 @@
namespace { // See SkNx.h
-template <>
-class SkNb<2, 4> {
-public:
- SkNb(const __m128i& vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return 0xff == (_mm_movemask_epi8(fVec) & 0xff); }
- bool anyTrue() const { return 0x00 != (_mm_movemask_epi8(fVec) & 0xff); }
-
- __m128i fVec;
-};
-
-template <>
-class SkNb<4, 4> {
-public:
- SkNb(const __m128i& vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return 0xffff == _mm_movemask_epi8(fVec); }
- bool anyTrue() const { return 0x0000 != _mm_movemask_epi8(fVec); }
-
- __m128i fVec;
-};
-
-template <>
-class SkNb<2, 8> {
-public:
- SkNb(const __m128i& vec) : fVec(vec) {}
-
- SkNb() {}
- bool allTrue() const { return 0xffff == _mm_movemask_epi8(fVec); }
- bool anyTrue() const { return 0x0000 != _mm_movemask_epi8(fVec); }
-
- __m128i fVec;
-};
-
template <>
class SkNf<2, float> {
- typedef SkNb<2, 4> Nb;
public:
SkNf(const __m128& vec) : fVec(vec) {}
@@ -69,12 +32,12 @@ public:
SkNf operator * (const SkNf& o) const { return _mm_mul_ps(fVec, o.fVec); }
SkNf operator / (const SkNf& o) const { return _mm_div_ps(fVec, o.fVec); }
- Nb operator == (const SkNf& o) const { return _mm_castps_si128(_mm_cmpeq_ps (fVec, o.fVec)); }
- Nb operator != (const SkNf& o) const { return _mm_castps_si128(_mm_cmpneq_ps(fVec, o.fVec)); }
- Nb operator < (const SkNf& o) const { return _mm_castps_si128(_mm_cmplt_ps (fVec, o.fVec)); }
- Nb operator > (const SkNf& o) const { return _mm_castps_si128(_mm_cmpgt_ps (fVec, o.fVec)); }
- Nb operator <= (const SkNf& o) const { return _mm_castps_si128(_mm_cmple_ps (fVec, o.fVec)); }
- Nb operator >= (const SkNf& o) const { return _mm_castps_si128(_mm_cmpge_ps (fVec, o.fVec)); }
+ SkNf operator == (const SkNf& o) const { return _mm_cmpeq_ps (fVec, o.fVec); }
+ SkNf operator != (const SkNf& o) const { return _mm_cmpneq_ps(fVec, o.fVec); }
+ SkNf operator < (const SkNf& o) const { return _mm_cmplt_ps (fVec, o.fVec); }
+ SkNf operator > (const SkNf& o) const { return _mm_cmpgt_ps (fVec, o.fVec); }
+ SkNf operator <= (const SkNf& o) const { return _mm_cmple_ps (fVec, o.fVec); }
+ SkNf operator >= (const SkNf& o) const { return _mm_cmpge_ps (fVec, o.fVec); }
static SkNf Min(const SkNf& l, const SkNf& r) { return _mm_min_ps(l.fVec, r.fVec); }
static SkNf Max(const SkNf& l, const SkNf& r) { return _mm_max_ps(l.fVec, r.fVec); }
@@ -93,12 +56,14 @@ public:
return pun.fs[k&1];
}
+ bool allTrue() const { return 0xff == (_mm_movemask_epi8(_mm_castps_si128(fVec)) & 0xff); }
+ bool anyTrue() const { return 0x00 != (_mm_movemask_epi8(_mm_castps_si128(fVec)) & 0xff); }
+
__m128 fVec;
};
template <>
class SkNf<2, double> {
- typedef SkNb<2, 8> Nb;
public:
SkNf(const __m128d& vec) : fVec(vec) {}
@@ -114,12 +79,12 @@ public:
SkNf operator * (const SkNf& o) const { return _mm_mul_pd(fVec, o.fVec); }
SkNf operator / (const SkNf& o) const { return _mm_div_pd(fVec, o.fVec); }
- Nb operator == (const SkNf& o) const { return _mm_castpd_si128(_mm_cmpeq_pd (fVec, o.fVec)); }
- Nb operator != (const SkNf& o) const { return _mm_castpd_si128(_mm_cmpneq_pd(fVec, o.fVec)); }
- Nb operator < (const SkNf& o) const { return _mm_castpd_si128(_mm_cmplt_pd (fVec, o.fVec)); }
- Nb operator > (const SkNf& o) const { return _mm_castpd_si128(_mm_cmpgt_pd (fVec, o.fVec)); }
- Nb operator <= (const SkNf& o) const { return _mm_castpd_si128(_mm_cmple_pd (fVec, o.fVec)); }
- Nb operator >= (const SkNf& o) const { return _mm_castpd_si128(_mm_cmpge_pd (fVec, o.fVec)); }
+ SkNf operator == (const SkNf& o) const { return _mm_cmpeq_pd (fVec, o.fVec); }
+ SkNf operator != (const SkNf& o) const { return _mm_cmpneq_pd(fVec, o.fVec); }
+ SkNf operator < (const SkNf& o) const { return _mm_cmplt_pd (fVec, o.fVec); }
+ SkNf operator > (const SkNf& o) const { return _mm_cmpgt_pd (fVec, o.fVec); }
+ SkNf operator <= (const SkNf& o) const { return _mm_cmple_pd (fVec, o.fVec); }
+ SkNf operator >= (const SkNf& o) const { return _mm_cmpge_pd (fVec, o.fVec); }
static SkNf Min(const SkNf& l, const SkNf& r) { return _mm_min_pd(l.fVec, r.fVec); }
static SkNf Max(const SkNf& l, const SkNf& r) { return _mm_max_pd(l.fVec, r.fVec); }
@@ -138,6 +103,9 @@ public:
return pun.ds[k&1];
}
+ bool allTrue() const { return 0xffff == _mm_movemask_epi8(_mm_castpd_si128(fVec)); }
+ bool anyTrue() const { return 0x0000 != _mm_movemask_epi8(_mm_castpd_si128(fVec)); }
+
__m128d fVec;
};
@@ -181,7 +149,6 @@ public:
template <>
class SkNf<4, float> {
- typedef SkNb<4, 4> Nb;
public:
SkNf(const __m128& vec) : fVec(vec) {}
@@ -199,12 +166,12 @@ public:
SkNf operator * (const SkNf& o) const { return _mm_mul_ps(fVec, o.fVec); }
SkNf operator / (const SkNf& o) const { return _mm_div_ps(fVec, o.fVec); }
- Nb operator == (const SkNf& o) const { return _mm_castps_si128(_mm_cmpeq_ps (fVec, o.fVec)); }
- Nb operator != (const SkNf& o) const { return _mm_castps_si128(_mm_cmpneq_ps(fVec, o.fVec)); }
- Nb operator < (const SkNf& o) const { return _mm_castps_si128(_mm_cmplt_ps (fVec, o.fVec)); }
- Nb operator > (const SkNf& o) const { return _mm_castps_si128(_mm_cmpgt_ps (fVec, o.fVec)); }
- Nb operator <= (const SkNf& o) const { return _mm_castps_si128(_mm_cmple_ps (fVec, o.fVec)); }
- Nb operator >= (const SkNf& o) const { return _mm_castps_si128(_mm_cmpge_ps (fVec, o.fVec)); }
+ SkNf operator == (const SkNf& o) const { return _mm_cmpeq_ps (fVec, o.fVec); }
+ SkNf operator != (const SkNf& o) const { return _mm_cmpneq_ps(fVec, o.fVec); }
+ SkNf operator < (const SkNf& o) const { return _mm_cmplt_ps (fVec, o.fVec); }
+ SkNf operator > (const SkNf& o) const { return _mm_cmpgt_ps (fVec, o.fVec); }
+ SkNf operator <= (const SkNf& o) const { return _mm_cmple_ps (fVec, o.fVec); }
+ SkNf operator >= (const SkNf& o) const { return _mm_cmpge_ps (fVec, o.fVec); }
static SkNf Min(const SkNf& l, const SkNf& r) { return _mm_min_ps(l.fVec, r.fVec); }
static SkNf Max(const SkNf& l, const SkNf& r) { return _mm_max_ps(l.fVec, r.fVec); }
@@ -223,6 +190,9 @@ public:
return pun.fs[k&3];
}
+ bool allTrue() const { return 0xffff == _mm_movemask_epi8(_mm_castps_si128(fVec)); }
+ bool anyTrue() const { return 0x0000 != _mm_movemask_epi8(_mm_castps_si128(fVec)); }
+
__m128 fVec;
};
@@ -312,6 +282,11 @@ public:
SkNi operator - (const SkNi& o) const { return _mm_sub_epi8(fVec, o.fVec); }
static SkNi Min(const SkNi& a, const SkNi& b) { return _mm_min_epu8(a.fVec, b.fVec); }
+ SkNi operator < (const SkNi& o) const {
+ // There's no unsigned _mm_cmplt_epu8, so we flip the sign bits then use a signed compare.
+ auto flip = _mm_set1_epi8(char(0x80));
+ return _mm_cmplt_epi8(_mm_xor_si128(flip, fVec), _mm_xor_si128(flip, o.fVec));
+ }
template <int k> uint8_t kth() const {
SkASSERT(0 <= k && k < 16);
@@ -320,6 +295,11 @@ public:
return k % 2 == 0 ? pair : (pair >> 8);
}
+ SkNi thenElse(const SkNi& t, const SkNi& e) const {
+ return _mm_or_si128(_mm_and_si128 (fVec, t.fVec),
+ _mm_andnot_si128(fVec, e.fVec));
+ }
+
__m128i fVec;
};
diff --git a/src/opts/SkXfermode_opts_SSE2.cpp b/src/opts/SkXfermode_opts_SSE2.cpp
index b92477094b..f8772808a7 100644
--- a/src/opts/SkXfermode_opts_SSE2.cpp
+++ b/src/opts/SkXfermode_opts_SSE2.cpp
@@ -515,15 +515,17 @@ void SkSSE2ProcCoeffXfermode::toString(SkString* str) const {
SkProcCoeffXfermode* SkPlatformXfermodeFactory_impl_SSE2(const ProcCoeff& rec,
SkXfermode::Mode mode) {
SkXfermodeProcSIMD proc = nullptr;
- // TODO(mtklein): implement these Sk4px.
switch (mode) {
+ // TODO(mtklein): Sk4pxXfermode has these now. Clean up.
case SkProcCoeffXfermode::kOverlay_Mode: proc = overlay_modeproc_SSE2; break;
case SkProcCoeffXfermode::kDarken_Mode: proc = darken_modeproc_SSE2; break;
case SkProcCoeffXfermode::kLighten_Mode: proc = lighten_modeproc_SSE2; break;
- case SkProcCoeffXfermode::kColorDodge_Mode: proc = colordodge_modeproc_SSE2; break;
- case SkProcCoeffXfermode::kColorBurn_Mode: proc = colorburn_modeproc_SSE2; break;
case SkProcCoeffXfermode::kHardLight_Mode: proc = hardlight_modeproc_SSE2; break;
+
+ // TODO(mtklein): implement these with SkPMFloat.
case SkProcCoeffXfermode::kSoftLight_Mode: proc = softlight_modeproc_SSE2; break;
+ case SkProcCoeffXfermode::kColorDodge_Mode: proc = colordodge_modeproc_SSE2; break;
+ case SkProcCoeffXfermode::kColorBurn_Mode: proc = colorburn_modeproc_SSE2; break;
default: break;
}
return proc ? SkNEW_ARGS(SkSSE2ProcCoeffXfermode, (rec, mode, (void*)proc)) : nullptr;
diff --git a/src/opts/SkXfermode_opts_arm_neon.cpp b/src/opts/SkXfermode_opts_arm_neon.cpp
index 1759429c57..205a00b4ce 100644
--- a/src/opts/SkXfermode_opts_arm_neon.cpp
+++ b/src/opts/SkXfermode_opts_arm_neon.cpp
@@ -1016,6 +1016,7 @@ SkProcCoeffXfermode* SkPlatformXfermodeFactory_impl_neon(const ProcCoeff& rec,
if (auto xfermode = SkCreate4pxXfermode(rec, mode)) {
return xfermode;
}
+ // TODO: Sk4pxXfermode now covers every mode found in this file. Delete them all!
if (auto proc = gNEONXfermodeProcs[mode]) {
return SkNEW_ARGS(SkNEONProcCoeffXfermode, (rec, mode, (void*)proc));
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-01 08:10:28 +0000