Implement Sk4i's abs, min, max

CQ_INCLUDE_TRYBOTS=skia.primary:Test-Debian9-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Bug: skia:
Change-Id: Ia9ec3f72095e1c744f88df7bb990d99e0f87d578
Reviewed-on: https://skia-review.googlesource.com/22720
Commit-Queue: Yuqian Li <liyuqian@google.com>
Reviewed-by: Herb Derby <herb@google.com>

2 files changed, 34 insertions, 0 deletions

diff --git a/src/opts/SkNx_neon.h b/src/opts/SkNx_neon.h
index 5671f71315..5ec3dc2f2b 100644
--- a/src/opts/SkNx_neon.h
+++ b/src/opts/SkNx_neon.h
@@ -433,12 +433,15 @@ public:
     }
 
     AI static SkNx Min(const SkNx& a, const SkNx& b) { return vminq_s32(a.fVec, b.fVec); }
+    AI static SkNx Max(const SkNx& a, const SkNx& b) { return vmaxq_s32(a.fVec, b.fVec); }
     // TODO as needed
 
     AI SkNx thenElse(const SkNx& t, const SkNx& e) const {
         return vbslq_s32(vreinterpretq_u32_s32(fVec), t.fVec, e.fVec);
     }
 
+    AI SkNx abs() const { return vabsq_s32(fVec); }
+
     int32x4_t fVec;
 };
 
diff --git a/src/opts/SkNx_sse.h b/src/opts/SkNx_sse.h
index 3e59a9d0e8..7542e3b4d7 100644
--- a/src/opts/SkNx_sse.h
+++ b/src/opts/SkNx_sse.h
@@ -198,6 +198,37 @@ public:
     #endif
     }
 
+    AI SkNx abs() const {
+#if SK_CPU_SSE_LEVEL >= SK_CPU_SSE_LEVEL_SSSE3
+        return _mm_abs_epi32(fVec);
+#else
+        SkNx mask = (*this) >> 31;
+        return (mask ^ (*this)) - mask;
+#endif
+    }
+
+    AI static SkNx Min(const SkNx& x, const SkNx& y) {
+#if SK_CPU_SSE_LEVEL >= SK_CPU_SSE_LEVEL_SSE41
+        return _mm_min_epi32(x.fVec, y.fVec);
+#else
+        __m128i less_than = _mm_cmplt_epi32(x.fVec, y.fVec);
+        __m128i choose_x = _mm_and_si128(less_than, x.fVec);
+        __m128i choose_y = _mm_andnot_si128(less_than, y.fVec);
+        return _mm_or_si128(choose_x, choose_y);
+#endif
+    }
+
+    AI static SkNx Max(const SkNx& x, const SkNx& y) {
+#if SK_CPU_SSE_LEVEL >= SK_CPU_SSE_LEVEL_SSE41
+        return _mm_max_epi32(x.fVec, y.fVec);
+#else
+        __m128i greater_than = _mm_cmpgt_epi32(x.fVec, y.fVec);
+        __m128i choose_x = _mm_and_si128(greater_than, x.fVec);
+        __m128i choose_y = _mm_andnot_si128(greater_than, y.fVec);
+        return _mm_or_si128(choose_x, choose_y);
+#endif
+    }
+
     __m128i fVec;
 };