This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [patch] malloc per-thread cache ready for review
- From: Markus Trippelsdorf <markus at trippelsdorf dot de>
- To: DJ Delorie <dj at redhat dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Wed, 1 Feb 2017 19:23:02 +0100
- Subject: Re: [patch] malloc per-thread cache ready for review
- Authentication-results: sourceware.org; auth=none
- References: <20170129094833.GA289@x4> <xnlgtql8t0.fsf@greed.delorie.com>
On 2017.02.01 at 01:33 -0500, DJ Delorie wrote:
>
> Markus Trippelsdorf <markus@trippelsdorf.de> writes:
> > Unfortunately it turned out to be quite unusable.
>
> Fixed. Turned out to be <= where I meant <
>
> Fix tested by building a full glibc rpm set, installing it, and building
> the full rpm set again using the just-installed glibc rpms.
>
> Please test again :-)
Thanks. It works fine now. But I see no speedups for a simple compile
time test:
% time g++ -w -Ofast tramp3d-v4.cpp
g++ -w -Ofast tramp3d-v4.cpp 26.42s user 0.31s system 99% cpu 26.756 total
(this is the same time as without your patch)
Using the Lockless Memory Allocator (http://locklessinc.com/articles/allocator_tricks/)
I get ~4% speedup:
% time LD_PRELOAD=/usr/lib/libllalloc.so.1.3 g++ -w -Ofast tramp3d-v4.cpp
LD_PRELOAD=/usr/lib/libllalloc.so.1.3 g++ -w -Ofast tramp3d-v4.cpp 25.25s user 0.38s system 99% cpu 25.659 total
--
Markus