blob: b05d2e9b1d6778c5560fc2b69f7a1de5d354e6d1 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
/* Compute the sum of the squares of a vector of signed shorts
* The SSE2 and MMX assist routines both operate on multiples of
* 8 words; they differ only in their alignment requirements (8 bytes
* for MMX, 16 bytes for SSE2)
* Copyright 2004 Phil Karn, KA9Q
* May be used under the terms of the GNU Lesser Public License (LGPL)
*/
long long sumsq_sse2_assist(signed short *,int);
long long sumsq_sse2(signed short *in,int cnt){
long long sum = 0;
/* Handle stuff before the next 8-byte boundary */
while(((int)in & 15) != 0 && cnt != 0){
sum += (long)in[0] * in[0];
in++;
cnt--;
}
sum += sumsq_sse2_assist(in,cnt);
in += cnt & ~7;
cnt &= 7;
/* Handle up to 7 trailing words */
while(cnt != 0){
sum += (long)in[0] * in[0];
in++;
cnt--;
}
return sum;
}
|
