Different processors has specific instructions to count leading
zero bits. This includes: x86. x64, arm, ppc.
For portability reason the behaviour of __builtin_clz is not
defined if the value is zero so test for it.
Currently the function is not called with the value or 0.
This increase performance decoding of about 4-5% on a x64 machine
(code size decreases a little too, but about 0.1%).
Signed-off-by: Frediano Ziglio <fziglio@xxxxxxxxxx>
---
common/quic.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/common/quic.c b/common/quic.c
index e31f789..8af826e 100644
--- a/common/quic.c
+++ b/common/quic.c
@@ -281,6 +281,12 @@ static const BYTE lzeroes[256] = {
/* count leading zeroes */
static unsigned int cnt_l_zeroes(const unsigned int bits)
{
+ if (spice_extra_checks) {
+ spice_assert(bits != 0);
+ }
+#if defined(__GNUC__) && __GNUC__ >= 4
+ return __builtin_clz(bits);
+#else
if (bits & 0xff800000) {
return lzeroes[bits >> 24];
} else if (bits & 0xffff8000) {
@@ -290,6 +296,7 @@ static unsigned int cnt_l_zeroes(const unsigned int bits)
} else {
return 24 + lzeroes[bits & 0x000000ff];
}
+#endif
}
#define QUIC_FAMILY_8BPC
--
2.17.1
_______________________________________________
Spice-devel mailing list
Spice-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/spice-devel