blob: 871ccc6a40c186a83a4d273021202bd11e4e3c9e (
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
34
35
36
37
38
39
|
--- src/arch/i386/x86_floatmul.c.orig 2005-12-21 22:06:35.000000000 +0300
+++ src/arch/i386/x86_floatmul.c 2009-04-04 04:53:56.000000000 +0400
@@ -46,7 +46,7 @@
v4hi temp;
- samples_pre = MMX_ALIGN - (aint)bpt % MMX_ALIGN;
+ samples_pre = (MMX_ALIGN - (aint)bpt % MMX_ALIGN) % MMX_ALIGN;
samples_pre /= sizeof(ALshort);
samples_main = len - samples_pre;
samples_post = samples_main % 8;
@@ -82,11 +82,27 @@ void _alFloatMul(ALshort *bpt, ALfloat s
v_sa.s[3] = v_sa.s[0];
while (samples_main--) {
+ union {
+ short s[4];
+ v4hi v;
+ } ALIGN16(v_one);
+ v_one.s[0] = 1;
+ v_one.s[1] = 0;
+ v_one.s[2] = 0;
+ v_one.s[3] = 0;
temp = __builtin_ia32_pmulhw(*(v4hi*)bpt, v_sa.v);
+#if defined(__GNUC__) && defined(__GNUC_MINOR__) && (__GNUC__ <= 3 || (__GNUC__ == 4 && __GNUC_MINOR__ <= 2)) && !defined(__clang__)
*(v4hi*)bpt = __builtin_ia32_psllw(temp, 1LL);
+#else
+ *(v4hi*)bpt = __builtin_ia32_psllw(temp, v_one.v);
+#endif
bpt += 4;
temp = __builtin_ia32_pmulhw(*(v4hi*)bpt, v_sa.v);
+#if defined(__GNUC__) && defined(__GNUC_MINOR__) && (__GNUC__ <= 3 || (__GNUC__ == 4 && __GNUC_MINOR__ <= 2)) && !defined(__clang__)
*(v4hi*)bpt = __builtin_ia32_psllw(temp, 1LL);
+#else
+ *(v4hi*)bpt = __builtin_ia32_psllw(temp, v_one.v);
+#endif
bpt += 4;
}
}
|
