X-Git-Url: https://git.zerfleddert.de/cgi-bin/gitweb.cgi/proxmark3-svn/blobdiff_plain/a4a6780efe814fd8ad96ffd85d01c1d71d3b7176..02515642bc81dc44d08e47b2f3454e29609dd0fe:/client/hardnested/hardnested_bitarray_core.c

diff --git a/client/hardnested/hardnested_bitarray_core.c b/client/hardnested/hardnested_bitarray_core.c
index c7e7f70f..aca4f149 100644
--- a/client/hardnested/hardnested_bitarray_core.c
+++ b/client/hardnested/hardnested_bitarray_core.c
@@ -21,17 +21,9 @@
 #include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
+#ifndef __APPLE__
 #include <malloc.h>
-
-// #include <stdint.h>
-// #include <stdbool.h>
-// #include <stdlib.h>
-// #include <stdio.h>
-// #include <malloc.h>
-// #include <string.h>
-// #include "crapto1/crapto1.h"
-// #include "parity.h"
-
+#endif
 
 // this needs to be compiled several times for each instruction set. 
 // For each instruction set, define a dedicated function name:
@@ -105,42 +97,63 @@
 #define COUNT_BITARRAY_AND2 count_bitarray_AND2_MMX
 #define COUNT_BITARRAY_AND3 count_bitarray_AND3_MMX
 #define COUNT_BITARRAY_AND4 count_bitarray_AND4_MMX
+#else
+#define MALLOC_BITARRAY malloc_bitarray_NOSIMD
+#define FREE_BITARRAY free_bitarray_NOSIMD
+#define BITCOUNT bitcount_NOSIMD
+#define COUNT_STATES count_states_NOSIMD
+#define BITARRAY_AND bitarray_AND_NOSIMD
+#define BITARRAY_LOW20_AND bitarray_low20_AND_NOSIMD
+#define COUNT_BITARRAY_AND count_bitarray_AND_NOSIMD
+#define COUNT_BITARRAY_LOW20_AND count_bitarray_low20_AND_NOSIMD
+#define BITARRAY_AND4 bitarray_AND4_NOSIMD
+#define BITARRAY_OR bitarray_OR_NOSIMD
+#define COUNT_BITARRAY_AND2 count_bitarray_AND2_NOSIMD
+#define COUNT_BITARRAY_AND3 count_bitarray_AND3_NOSIMD
+#define COUNT_BITARRAY_AND4 count_bitarray_AND4_NOSIMD
 #endif
 
 
 // typedefs and declaration of functions:
 typedef uint32_t* malloc_bitarray_t(uint32_t);
-malloc_bitarray_t malloc_bitarray_AVX512, malloc_bitarray_AVX2, malloc_bitarray_AVX, malloc_bitarray_SSE2, malloc_bitarray_MMX, malloc_bitarray_dispatch;
+malloc_bitarray_t malloc_bitarray_AVX512, malloc_bitarray_AVX2, malloc_bitarray_AVX, malloc_bitarray_SSE2, malloc_bitarray_MMX, malloc_bitarray_NOSIMD, malloc_bitarray_dispatch;
 typedef void free_bitarray_t(uint32_t*);
-free_bitarray_t free_bitarray_AVX512, free_bitarray_AVX2, free_bitarray_AVX, free_bitarray_SSE2, free_bitarray_MMX, free_bitarray_dispatch;
+free_bitarray_t free_bitarray_AVX512, free_bitarray_AVX2, free_bitarray_AVX, free_bitarray_SSE2, free_bitarray_MMX, free_bitarray_NOSIMD, free_bitarray_dispatch;
 typedef uint32_t bitcount_t(uint32_t);
-bitcount_t bitcount_AVX512, bitcount_AVX2, bitcount_AVX, bitcount_SSE2, bitcount_MMX, bitcount_dispatch;
+bitcount_t bitcount_AVX512, bitcount_AVX2, bitcount_AVX, bitcount_SSE2, bitcount_MMX, bitcount_NOSIMD, bitcount_dispatch;
 typedef uint32_t count_states_t(uint32_t*);
-count_states_t count_states_AVX512, count_states_AVX2, count_states_AVX, count_states_SSE2, count_states_MMX, count_states_dispatch;
+count_states_t count_states_AVX512, count_states_AVX2, count_states_AVX, count_states_SSE2, count_states_MMX, count_states_NOSIMD, count_states_dispatch;
 typedef void bitarray_AND_t(uint32_t[], uint32_t[]);
-bitarray_AND_t bitarray_AND_AVX512, bitarray_AND_AVX2, bitarray_AND_AVX, bitarray_AND_SSE2, bitarray_AND_MMX, bitarray_AND_dispatch;
+bitarray_AND_t bitarray_AND_AVX512, bitarray_AND_AVX2, bitarray_AND_AVX, bitarray_AND_SSE2, bitarray_AND_MMX, bitarray_AND_NOSIMD, bitarray_AND_dispatch;
 typedef void bitarray_low20_AND_t(uint32_t*, uint32_t*);
-bitarray_low20_AND_t bitarray_low20_AND_AVX512, bitarray_low20_AND_AVX2, bitarray_low20_AND_AVX, bitarray_low20_AND_SSE2, bitarray_low20_AND_MMX, bitarray_low20_AND_dispatch;
+bitarray_low20_AND_t bitarray_low20_AND_AVX512, bitarray_low20_AND_AVX2, bitarray_low20_AND_AVX, bitarray_low20_AND_SSE2, bitarray_low20_AND_MMX, bitarray_low20_AND_NOSIMD, bitarray_low20_AND_dispatch;
 typedef uint32_t count_bitarray_AND_t(uint32_t*, uint32_t*);
-count_bitarray_AND_t count_bitarray_AND_AVX512, count_bitarray_AND_AVX2, count_bitarray_AND_AVX, count_bitarray_AND_SSE2, count_bitarray_AND_MMX, count_bitarray_AND_dispatch;
+count_bitarray_AND_t count_bitarray_AND_AVX512, count_bitarray_AND_AVX2, count_bitarray_AND_AVX, count_bitarray_AND_SSE2, count_bitarray_AND_MMX, count_bitarray_AND_NOSIMD, count_bitarray_AND_dispatch;
 typedef uint32_t count_bitarray_low20_AND_t(uint32_t*, uint32_t*);
-count_bitarray_low20_AND_t count_bitarray_low20_AND_AVX512, count_bitarray_low20_AND_AVX2, count_bitarray_low20_AND_AVX, count_bitarray_low20_AND_SSE2, count_bitarray_low20_AND_MMX, count_bitarray_low20_AND_dispatch;
+count_bitarray_low20_AND_t count_bitarray_low20_AND_AVX512, count_bitarray_low20_AND_AVX2, count_bitarray_low20_AND_AVX, count_bitarray_low20_AND_SSE2, count_bitarray_low20_AND_MMX, count_bitarray_low20_AND_NOSIMD, count_bitarray_low20_AND_dispatch;
 typedef void bitarray_AND4_t(uint32_t*, uint32_t*, uint32_t*, uint32_t*);
-bitarray_AND4_t bitarray_AND4_AVX512, bitarray_AND4_AVX2, bitarray_AND4_AVX, bitarray_AND4_SSE2, bitarray_AND4_MMX, bitarray_AND4_dispatch;
+bitarray_AND4_t bitarray_AND4_AVX512, bitarray_AND4_AVX2, bitarray_AND4_AVX, bitarray_AND4_SSE2, bitarray_AND4_MMX, bitarray_AND4_NOSIMD, bitarray_AND4_dispatch;
 typedef void bitarray_OR_t(uint32_t[], uint32_t[]);
-bitarray_OR_t bitarray_OR_AVX512, bitarray_OR_AVX2, bitarray_OR_AVX, bitarray_OR_SSE2, bitarray_OR_MMX, bitarray_OR_dispatch;
+bitarray_OR_t bitarray_OR_AVX512, bitarray_OR_AVX2, bitarray_OR_AVX, bitarray_OR_SSE2, bitarray_OR_MMX, bitarray_OR_NOSIMD, bitarray_OR_dispatch;
 typedef uint32_t count_bitarray_AND2_t(uint32_t*, uint32_t*);
-count_bitarray_AND2_t count_bitarray_AND2_AVX512, count_bitarray_AND2_AVX2, count_bitarray_AND2_AVX, count_bitarray_AND2_SSE2, count_bitarray_AND2_MMX, count_bitarray_AND2_dispatch;
+count_bitarray_AND2_t count_bitarray_AND2_AVX512, count_bitarray_AND2_AVX2, count_bitarray_AND2_AVX, count_bitarray_AND2_SSE2, count_bitarray_AND2_MMX, count_bitarray_AND2_NOSIMD, count_bitarray_AND2_dispatch;
 typedef uint32_t count_bitarray_AND3_t(uint32_t*, uint32_t*, uint32_t*);
-count_bitarray_AND3_t count_bitarray_AND3_AVX512, count_bitarray_AND3_AVX2, count_bitarray_AND3_AVX, count_bitarray_AND3_SSE2, count_bitarray_AND3_MMX, count_bitarray_AND3_dispatch;
+count_bitarray_AND3_t count_bitarray_AND3_AVX512, count_bitarray_AND3_AVX2, count_bitarray_AND3_AVX, count_bitarray_AND3_SSE2, count_bitarray_AND3_MMX, count_bitarray_AND3_NOSIMD, count_bitarray_AND3_dispatch;
 typedef uint32_t count_bitarray_AND4_t(uint32_t*, uint32_t*, uint32_t*, uint32_t*);
-count_bitarray_AND4_t count_bitarray_AND4_AVX512, count_bitarray_AND4_AVX2, count_bitarray_AND4_AVX, count_bitarray_AND4_SSE2, count_bitarray_AND4_MMX, count_bitarray_AND4_dispatch;
+count_bitarray_AND4_t count_bitarray_AND4_AVX512, count_bitarray_AND4_AVX2, count_bitarray_AND4_AVX, count_bitarray_AND4_SSE2, count_bitarray_AND4_MMX, count_bitarray_AND4_NOSIMD, count_bitarray_AND4_dispatch;
 
 
 inline uint32_t *MALLOC_BITARRAY(uint32_t x)
 {
-#ifdef _WIN32
+#if defined (_WIN32)
 	return __builtin_assume_aligned(_aligned_malloc((x), __BIGGEST_ALIGNMENT__), __BIGGEST_ALIGNMENT__);
+#elif defined (__APPLE__)
+	uint32_t *allocated_memory;
+	if (posix_memalign((void**)&allocated_memory, __BIGGEST_ALIGNMENT__, x)) {
+		return NULL;
+	} else {
+		return __builtin_assume_aligned(allocated_memory, __BIGGEST_ALIGNMENT__);
+	}
 #else
 	return __builtin_assume_aligned(memalign(__BIGGEST_ALIGNMENT__, (x)), __BIGGEST_ALIGNMENT__);
 #endif
@@ -285,6 +298,7 @@ inline uint32_t COUNT_BITARRAY_AND4(uint32_t *restrict A, uint32_t *restrict B,
 	return count;
 }
 
+
 #ifndef __MMX__
 
 // pointers to functions:
@@ -304,235 +318,274 @@ count_bitarray_AND4_t *count_bitarray_AND4_function_p = &count_bitarray_AND4_dis
 
 // determine the available instruction set at runtime and call the correct function
 uint32_t *malloc_bitarray_dispatch(uint32_t x) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) malloc_bitarray_function_p = &malloc_bitarray_AVX512;
 	else if (__builtin_cpu_supports("avx2")) malloc_bitarray_function_p = &malloc_bitarray_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) malloc_bitarray_function_p = &malloc_bitarray_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) malloc_bitarray_function_p = &malloc_bitarray_AVX;
 	else if (__builtin_cpu_supports("sse2")) malloc_bitarray_function_p = &malloc_bitarray_SSE2;
 	else if (__builtin_cpu_supports("mmx")) malloc_bitarray_function_p = &malloc_bitarray_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif		
+		malloc_bitarray_function_p = &malloc_bitarray_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*malloc_bitarray_function_p)(x);
 }
 
 void free_bitarray_dispatch(uint32_t *x) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) free_bitarray_function_p = &free_bitarray_AVX512;
 	else if (__builtin_cpu_supports("avx2")) free_bitarray_function_p = &free_bitarray_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) free_bitarray_function_p = &free_bitarray_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) free_bitarray_function_p = &free_bitarray_AVX;
 	else if (__builtin_cpu_supports("sse2")) free_bitarray_function_p = &free_bitarray_SSE2;
 	else if (__builtin_cpu_supports("mmx")) free_bitarray_function_p = &free_bitarray_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		free_bitarray_function_p = &free_bitarray_NOSIMD;
+
     // call the most optimized function for this CPU
     (*free_bitarray_function_p)(x);
 }
 
 uint32_t bitcount_dispatch(uint32_t a) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) bitcount_function_p = &bitcount_AVX512;
 	else if (__builtin_cpu_supports("avx2")) bitcount_function_p = &bitcount_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) bitcount_function_p = &bitcount_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) bitcount_function_p = &bitcount_AVX;
 	else if (__builtin_cpu_supports("sse2")) bitcount_function_p = &bitcount_SSE2;
 	else if (__builtin_cpu_supports("mmx")) bitcount_function_p = &bitcount_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		bitcount_function_p = &bitcount_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*bitcount_function_p)(a);
 }
 
 uint32_t count_states_dispatch(uint32_t *bitarray) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_states_function_p = &count_states_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_states_function_p = &count_states_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_states_function_p = &count_states_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_states_function_p = &count_states_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_states_function_p = &count_states_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_states_function_p = &count_states_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif 
+#endif
+		count_states_function_p = &count_states_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_states_function_p)(bitarray);
 }
 
 void bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) bitarray_AND_function_p = &bitarray_AND_AVX512;
 	else if (__builtin_cpu_supports("avx2")) bitarray_AND_function_p = &bitarray_AND_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) bitarray_AND_function_p = &bitarray_AND_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) bitarray_AND_function_p = &bitarray_AND_AVX;
 	else if (__builtin_cpu_supports("sse2")) bitarray_AND_function_p = &bitarray_AND_SSE2;
 	else if (__builtin_cpu_supports("mmx")) bitarray_AND_function_p = &bitarray_AND_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		bitarray_AND_function_p = &bitarray_AND_NOSIMD;
+
     // call the most optimized function for this CPU
     (*bitarray_AND_function_p)(A,B);
 }
 
 void bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX512;
 	else if (__builtin_cpu_supports("avx2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX;
 	else if (__builtin_cpu_supports("sse2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_SSE2;
 	else if (__builtin_cpu_supports("mmx")) bitarray_low20_AND_function_p = &bitarray_low20_AND_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		bitarray_low20_AND_function_p = &bitarray_low20_AND_NOSIMD;
+
     // call the most optimized function for this CPU
     (*bitarray_low20_AND_function_p)(A, B);
 }
 
 uint32_t count_bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_bitarray_AND_function_p = &count_bitarray_AND_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_bitarray_AND_function_p = &count_bitarray_AND_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		count_bitarray_AND_function_p = &count_bitarray_AND_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_bitarray_AND_function_p)(A, B);
 }
 
 uint32_t count_bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_bitarray_low20_AND_function_p)(A, B);
 }
 
 void bitarray_AND4_dispatch(uint32_t *A, uint32_t *B, uint32_t *C, uint32_t *D) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) bitarray_AND4_function_p = &bitarray_AND4_AVX512;
 	else if (__builtin_cpu_supports("avx2")) bitarray_AND4_function_p = &bitarray_AND4_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) bitarray_AND4_function_p = &bitarray_AND4_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) bitarray_AND4_function_p = &bitarray_AND4_AVX;
 	else if (__builtin_cpu_supports("sse2")) bitarray_AND4_function_p = &bitarray_AND4_SSE2;
 	else if (__builtin_cpu_supports("mmx")) bitarray_AND4_function_p = &bitarray_AND4_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		bitarray_AND4_function_p = &bitarray_AND4_NOSIMD;
+
     // call the most optimized function for this CPU
     (*bitarray_AND4_function_p)(A, B, C, D);
 }
 
 void bitarray_OR_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) bitarray_OR_function_p = &bitarray_OR_AVX512;
 	else if (__builtin_cpu_supports("avx2")) bitarray_OR_function_p = &bitarray_OR_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) bitarray_OR_function_p = &bitarray_OR_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) bitarray_OR_function_p = &bitarray_OR_AVX;
 	else if (__builtin_cpu_supports("sse2")) bitarray_OR_function_p = &bitarray_OR_SSE2;
 	else if (__builtin_cpu_supports("mmx")) bitarray_OR_function_p = &bitarray_OR_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		bitarray_OR_function_p = &bitarray_OR_NOSIMD;
+
     // call the most optimized function for this CPU
     (*bitarray_OR_function_p)(A,B);
 }
 
 uint32_t count_bitarray_AND2_dispatch(uint32_t *A, uint32_t *B) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_bitarray_AND2_function_p = &count_bitarray_AND2_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		count_bitarray_AND2_function_p = &count_bitarray_AND2_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_bitarray_AND2_function_p)(A, B);
 }
 
 uint32_t count_bitarray_AND3_dispatch(uint32_t *A, uint32_t *B, uint32_t *C) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_bitarray_AND3_function_p = &count_bitarray_AND3_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		count_bitarray_AND3_function_p = &count_bitarray_AND3_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_bitarray_AND3_function_p)(A, B, C);
 }
 
 uint32_t count_bitarray_AND4_dispatch(uint32_t *A, uint32_t *B, uint32_t *C, uint32_t *D) {
-	#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)	
+	#if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8 || __clang_major__ == 8 && __clang_minor__ >= 1))
+		#if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
 	if (__builtin_cpu_supports("avx512f")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX512;
 	else if (__builtin_cpu_supports("avx2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX2;
-	#else
+		#else
 	if (__builtin_cpu_supports("avx2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX2;
-	#endif
+		#endif
 	else if (__builtin_cpu_supports("avx")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX;
 	else if (__builtin_cpu_supports("sse2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_SSE2;
 	else if (__builtin_cpu_supports("mmx")) count_bitarray_AND4_function_p = &count_bitarray_AND4_MMX;
-    else {
-        printf("\nFatal: you need at least a CPU with MMX instruction set support. Aborting...\n");
-        exit(5);
-    }
+	else
+	#endif
+#endif
+		count_bitarray_AND4_function_p = &count_bitarray_AND4_NOSIMD;
+
     // call the most optimized function for this CPU
     return (*count_bitarray_AND4_function_p)(A, B, C, D);
 }