Henrik Nordstrom wrote:
> Quick check.. cond_signal of an empty cond is a rather heavy operation
> and should be avoided if possible. Changes things a little.
Turned out to be not as heavy as first measurements indicate..
P133, Linux-2.2, glibc-2.1.3
cond_signal 1.2us
write(fd, "x", 1) 12us
PIII 1GHz, Linux-2.4.18-SMP (but only 1 CPU installed), glibc-2.2.4
cond_signal 0.14us
write 14us
but still most likely heavier than a simple counter filter within the
already used mutex..
Regards
Henrik
Received on Sat Jan 04 2003 - 16:31:27 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:19:05 MST