[sldev] [VWR] Heap checkers

Robin Cornelius robin.cornelius at gmail.com
Thu Jul 31 10:23:17 PDT 2008


Bryan O'Sullivan wrote:
> On Thu, Jul 31, 2008 at 9:53 AM, Brad Kittenbrink (Brad Linden)
> <brad at lindenlab.com> wrote:
> 
>> Internally we've been investigating using tcmalloc on windows.  Our tests
>> with heap checking enabled have resulted in framerates of about 4-5fps once
>> logged in.  We haven't focused as much on non-windows platforms yet, frankly
>> because replacing the allocator is likely easier there.  From what I recall,
>> we didn't run into any problems with tcmalloc on linux / OS X, but I'll
>> double check on that.  I'm very curious about the gtk thread issue you
>> mention.
> 
> We used to build and ship the Linux viewer against tcmalloc, but we
> gave up because of reports of crashes that we were unable to
> reproduce. Perhaps the situation has since improved.
> 

On the 1.20.15 branch tcmalloc seems to behave perfectly and certainly
does not introduce any instability of its own (in normal, not debug mode)

Its suppose to have a slightly faster malloc() something like 150us than
glibc (on some mid range setup) but in the grand picture of things that
is probably insignificant. So apart from when you want to explicitly
turn on heap checking or heap profiling i don't see any amazing
advantage for normal operation and if you are getting this deep, linking
in your own debug malloc such as tcmalloc is not an issue.

Robin



-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 260 bytes
Desc: OpenPGP digital signature
Url : http://lists.secondlife.com/pipermail/sldev/attachments/20080731/0a1baedf/signature.pgp


More information about the SLDev mailing list