}
+ALIGN_STACK
static boolean
test_one(unsigned verbose,
FILE *fp,
success = TRUE;
for(i = 0; i < n && success; ++i) {
if(mode == AoS) {
- uint8_t src[LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t dst[LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t con[LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t res[LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t ref[LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t src[LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t dst[LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t con[LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t res[LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t ref[LP_NATIVE_VECTOR_WIDTH/8];
int64_t start_counter = 0;
int64_t end_counter = 0;
if(mode == SoA) {
const unsigned stride = type.length*type.width/8;
- uint8_t src[4*LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t dst[4*LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t con[4*LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t res[4*LP_NATIVE_VECTOR_WIDTH/8];
- uint8_t ref[4*LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t src[4*LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t dst[4*LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t con[4*LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t res[4*LP_NATIVE_VECTOR_WIDTH/8];
+ ALIGN16_ATTRIB uint8_t ref[4*LP_NATIVE_VECTOR_WIDTH/8];
int64_t start_counter = 0;
int64_t end_counter = 0;
boolean mismatch;
}
+ALIGN_STACK
static boolean
test_one(unsigned verbose,
FILE *fp,
for(i = 0; i < n && success; ++i) {
unsigned src_stride = src_type.length*src_type.width/8;
unsigned dst_stride = dst_type.length*dst_type.width/8;
- uint8_t src[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
- uint8_t dst[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
+ ALIGN16_ATTRIB uint8_t src[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
+ ALIGN16_ATTRIB uint8_t dst[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
double fref[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
uint8_t ref[LP_MAX_VECTOR_LENGTH*LP_MAX_VECTOR_LENGTH];
int64_t start_counter = 0;