ARM: optimised MUL64
Originally committed as revision 14244 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
parent
7995962d4b
commit
6865e04620
@ -48,6 +48,15 @@ static inline av_const int MULH(int a, int b)
|
||||
hi; })
|
||||
#endif
|
||||
|
||||
static inline av_const int64_t MUL64(int a, int b)
|
||||
{
|
||||
union { uint64_t x; unsigned hl[2]; } x;
|
||||
asm ("smull %0, %1, %2, %3"
|
||||
: "=r"(x.hl[0]), "=r"(x.hl[1]) : "r"(a), "r"(b));
|
||||
return x.x;
|
||||
}
|
||||
#define MUL64 MUL64
|
||||
|
||||
#if defined(HAVE_ARMV5TE)
|
||||
|
||||
/* signed 16x16 -> 32 multiply add accumulate */
|
||||
|
Loading…
Reference in New Issue
Block a user