Hi all,
For my own curiosity regarding the overhead with memory-limit or keeping
track of usage so memory_get_[peak_]usage() can always be enabled, I just
did some quick testing. Using this code
PHP_FUNCTION(emalloc_tester)
{
int i;
long mem_size;
void *ptr;
if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "l", &mem_size) ==
FAILURE) {
return;
}
for (i = 0; i < 100000; i++) {
ptr = emalloc(mem_size);
efree(ptr);
}
}
with different mem_size's and timing it. Using the latest CVS files for
5.2; only difference is the --enable-memory-limit option. Dual 930MHz PIII
Windows 2000 system (yes, *both* CPUs were maxed after mem_size got larger
than would fit into a block :-)).
-ms-
Size On Off
------------------
8 10.6 10.8
32 10.7 10.8
128 10.6 10.8
512 16.2 16.5
1k 16.3 16.5
8k 16.3 16.6
64k 16.3 16.6
128k 16.3 16.6
256k 2155 2140
512k 2225 2210
Interesting that it was consistently faster for me *with* memory-limit at
the smaller sizes, which I assume are by far the most-used. Does this help
at all? :-)
Matt
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
