vppinfra: add vector horizontal add and byte swap (SSE4.2 & AVX2)
Change-Id: I4e0fd487970796f0153a5b16333827d23b57deac
Signed-off-by: Damjan Marion <damarion@cisco.com>
diff --git a/src/vppinfra/vector_sse42.h b/src/vppinfra/vector_sse42.h
index fa377cc..deb9877 100644
--- a/src/vppinfra/vector_sse42.h
+++ b/src/vppinfra/vector_sse42.h
@@ -592,6 +592,21 @@
#undef _signed_binop
+static_always_inline u16x8
+u16x8_byte_swap (u16x8 v)
+{
+ u8x16 swap = {
+ 1, 0, 3, 2, 5, 4, 7, 6, 9, 8, 11, 10, 13, 12, 15, 14,
+ };
+ return (u16x8) _mm_shuffle_epi8 ((__m128i) v, (__m128i) swap);
+}
+
+static_always_inline u32x4
+u32x4_hadd (u32x4 v1, u32x4 v2)
+{
+ return (u32x4) _mm_hadd_epi32 ((__m128i) v1, (__m128i) v2);
+}
+
#endif /* included_vector_sse2_h */
/*