Re: [Yusuf Goolamabbas <yusufg@outblaze.com>] [dgaudet@arctic.org: faster malloc for multithreaded programs]
SL Baur writes:
Does someone want to take a look at this?
---------- Forwarded message ----------
From: "Emery Berger" <emery(a)cs.utexas.edu>
Subject: Hoard 1.2 - malloc/free replacement: 30% faster + SPROC support
Date: Sun, 31 Oct 1999 08:43:23 GMT
=====BEGIN PGP SIGNED MESSAGE=====
Hoard is a fast, scalable, and memory-efficient replacement
for malloc/free for multiprocessors. The newest release (1.2)
incorporates a user-level lock implementation (for x86, SPARC,
and SGI's) that improves performance by about 30%. This makes
Hoard faster than the Linux allocator in glibc2, even on just
one processor (it's always been faster for multiple processors).
This version also includes an "sproc" version for SGI's.
<http://www.cs.utexas.edu/users/emery/hoard>
Looks like some work has to be done for AIX, HP-UX and others. Anyway,
would be interesting to know how this compares to other implementation
on e.g. AIX/HP-UX.