| 1 |
Author: Michael R. Crusoe <[email protected]> |
| 2 |
Description: Enable SIMDE everywhere |
| 3 |
Forwarded: not-needed |
| 4 |
--- a/sse_wrap.h |
| 5 |
+++ b/sse_wrap.h |
| 6 |
@@ -27,8 +27,9 @@ |
| 7 |
#ifndef SSE_WRAP_H_ |
| 8 |
#define SSE_WRAP_H_ |
| 9 |
|
| 10 |
+#define SIMDE_ENABLE_NATIVE_ALIASES |
| 11 |
#ifdef SSE_AVX2 |
| 12 |
-#include <immintrin.h> |
| 13 |
+#include <simde/x86/avx2.h> |
| 14 |
#define NBYTES_PER_REG 32 |
| 15 |
#define BYTES_LOG2_PER_REG 5 |
| 16 |
#define SSE_MASK_ALL ((int) 0xffffffff) |
| 17 |
@@ -109,49 +110,13 @@ |
| 18 |
outval = outval & 0x00ff; \ |
| 19 |
} |
| 20 |
|
| 21 |
- |
| 22 |
#else /* no SSE_AVX2 */ |
| 23 |
|
| 24 |
-#if defined(__aarch64__) || defined(__s390x__) || defined(__powerpc__) |
| 25 |
-#include "simde/x86/sse2.h" |
| 26 |
-#else |
| 27 |
-#include <emmintrin.h> |
| 28 |
-#endif |
| 29 |
- |
| 30 |
+#include <simde/x86/sse2.h> |
| 31 |
#define NBYTES_PER_REG 16 |
| 32 |
#define BYTES_LOG2_PER_REG 4 |
| 33 |
#define SSE_MASK_ALL 0xffff |
| 34 |
|
| 35 |
-#if defined(__aarch64__) || defined(__s390x__) || defined(__powerpc__) |
| 36 |
-typedef simde__m128i SSERegI; |
| 37 |
-#define sse_adds_epi16(x, y) simde_mm_adds_epi16(x, y) |
| 38 |
-#define sse_adds_epu8(x, y) simde_mm_adds_epu8(x, y) |
| 39 |
-#define sse_cmpeq_epi16(x, y) simde_mm_cmpeq_epi16(x, y) |
| 40 |
-#define sse_cmpeq_epi8(x, y) simde_mm_cmpeq_epi8(x, y) |
| 41 |
-#define sse_cmpgt_epi16(x, y) simde_mm_cmpgt_epi16(x, y) |
| 42 |
-#define sse_cmpgt_epi8(x, y) simde_mm_cmpgt_epi8(x, y) |
| 43 |
-#define sse_cmplt_epi16(x, y) simde_mm_cmplt_epi16(x, y) |
| 44 |
-#define sse_cmplt_epu8(x, y) simde_mm_cmplt_epu8(x, y) |
| 45 |
-#define sse_extract_epi16(x, y) simde_mm_extract_epi16(x, y) |
| 46 |
-#define sse_insert_epi16(x, y, z) simde_mm_insert_epi16(x, y, z) |
| 47 |
-#define sse_load_siall(x) simde_mm_load_si128(x) |
| 48 |
-#define sse_max_epi16(x, y) simde_mm_max_epi16(x, y) |
| 49 |
-#define sse_max_epu8(x, y) simde_mm_max_epu8(x, y) |
| 50 |
-#define sse_movemask_epi8(x) simde_mm_movemask_epi8(x) |
| 51 |
-#define sse_or_siall(x, y) simde_mm_or_si128(x, y) |
| 52 |
-#define sse_setzero_siall() simde_mm_setzero_si128() |
| 53 |
-#define sse_slli_epi16(x, y) simde_mm_slli_epi16(x, y) |
| 54 |
-#define sse_slli_siall(x, y) simde_mm_slli_si128(x, y) |
| 55 |
-#define sse_srli_epi16(x, y) simde_mm_srli_epi16(x, y) |
| 56 |
-#define sse_srli_epu8(x, y) simde_mm_srli_epu8(x, y) |
| 57 |
-#define sse_srli_siall(x, y) simde_mm_srli_si128(x, y) |
| 58 |
-#define sse_store_siall(x, y) simde_mm_store_si128(x, y) |
| 59 |
-#define sse_subs_epi16(x, y) simde_mm_subs_epi16(x, y) |
| 60 |
-#define sse_subs_epu8(x, y) simde_mm_subs_epu8(x, y) |
| 61 |
-#define sse_xor_siall(x, y) simde_mm_xor_si128(x, y) |
| 62 |
-#define sse_set1_epi16(x) simde_mm_set1_epi16(x) |
| 63 |
- |
| 64 |
-#else |
| 65 |
typedef __m128i SSERegI; |
| 66 |
#define sse_adds_epi16(x, y) _mm_adds_epi16(x, y) |
| 67 |
#define sse_adds_epu8(x, y) _mm_adds_epu8(x, y) |
| 68 |
@@ -180,8 +145,6 @@ |
| 69 |
#define sse_xor_siall(x, y) _mm_xor_si128(x, y) |
| 70 |
#define sse_set1_epi16(x) _mm_set1_epi16(x) |
| 71 |
|
| 72 |
-#endif |
| 73 |
- |
| 74 |
/* compute the max val of a vector */ |
| 75 |
#define sse_max_score_i16(inval, outval) { \ |
| 76 |
SSERegI vlmax = inval; \ |