sse: added NEON impl for mm_shuffle_ps

3d4512c4 · Hidayat Khan · Michael R. Crusoe · c78ae2f5 · 3d4512c4 · 3d4512c4
Commit 3d4512c4 authored Dec 19, 2020 by Hidayat Khan Committed by Michael R. Crusoe Dec 21, 2020
Hide whitespace changes
Inline Side-by-side

Showing with 17 additions and 1 deletion

x86/sse.h x86/sse.h +16 -0

x86/sse2.h x86/sse2.h +1 -1

No files found.
--- a/x86/sse.h
+++ b/x86/sse.h
@@ -3579,6 +3579,22 @@ HEDLEY_DIAGNOSTIC_POP
 #if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
 #  define simde_mm_shuffle_ps(a, b, imm8) _mm_shuffle_ps(a, b, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+  #define simde_mm_shuffle_ps(a, b, imm8)                                   \
+     __extension__({                                                        \
+        float32x4_t ret;                                                   \
+        ret = vmovq_n_f32(                                                 \
+            vgetq_lane_f32(a, (imm8) & (0x3)));     \
+        ret = vsetq_lane_f32(                                              \
+            vgetq_lane_f32(a, ((imm8) >> 2) & 0x3), \
+            ret, 1);                                                       \
+        ret = vsetq_lane_f32(                                              \
+            vgetq_lane_f32(b, ((imm8) >> 4) & 0x3), \
+            ret, 2);                                                       \
+        ret = vsetq_lane_f32(                                              \
+            vgetq_lane_f32(b, ((imm8) >> 6) & 0x3), \
+            ret, 3);                                                                    \
+    })
 #elif defined(SIMDE_SHUFFLE_VECTOR_)
 #  define simde_mm_shuffle_ps(a, b, imm8) (__extension__ ({ \
      simde__m128_from_private((simde__m128_private) { .f32 = \

--- a/x86/sse2.h
+++ b/x86/sse2.h
@@ -5097,7 +5097,7 @@ simde_mm_shuffle_epi32 (simde__m128i a, const int imm8)
 #if defined(SIMDE_X86_SSE2_NATIVE)
  #define simde_mm_shuffle_epi32(a, imm8) _mm_shuffle_epi32((a), (imm8))
 #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-  #define _mm_shuffle_epi32(a, imm8)                                   \
+  #define simde_mm_shuffle_epi32(a, imm8)                                   \
    __extension__({                                                         \
        int32x4_t ret;                                                      \
        ret = vmovq_n_s32(                                                  \