@@ -28,7 +28,7 @@ double DotProductSSE(const double* u, const double* v, int n) {
28
28
fprintf (stderr, " DotProductSSE can't be used on Android\n " );
29
29
abort ();
30
30
}
31
- inT32 IntDotProductSSE (const inT8 * u, const inT8 * v, int n) {
31
+ int32_t IntDotProductSSE (const int8_t * u, const int8_t * v, int n) {
32
32
fprintf (stderr, " IntDotProductSSE can't be used on Android\n " );
33
33
abort ();
34
34
}
@@ -99,7 +99,7 @@ double DotProductSSE(const double* u, const double* v, int n) {
99
99
100
100
// Computes and returns the dot product of the n-vectors u and v.
101
101
// Uses Intel SSE intrinsics to access the SIMD instruction set.
102
- inT32 IntDotProductSSE (const inT8 * u, const inT8 * v, int n) {
102
+ int32_t IntDotProductSSE (const int8_t * u, const int8_t * v, int n) {
103
103
int max_offset = n - 8 ;
104
104
int offset = 0 ;
105
105
// Accumulate a set of 4 32-bit sums in sum, by loading 8 pairs of 8-bit
@@ -128,7 +128,7 @@ inT32 IntDotProductSSE(const inT8* u, const inT8* v, int n) {
128
128
// Sum the 4 packed 32 bit sums and extract the low result.
129
129
sum = _mm_hadd_epi32 (sum, sum);
130
130
sum = _mm_hadd_epi32 (sum, sum);
131
- inT32 result = _mm_cvtsi128_si32 (sum);
131
+ int32_t result = _mm_cvtsi128_si32 (sum);
132
132
while (offset < n) {
133
133
result += u[offset] * v[offset];
134
134
++offset;
0 commit comments