Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 10 Dec 2015 14:43:50 +0100
From: Szabolcs Nagy <>
Cc: Ed Schouten <>
Subject: Re: Re: AVL tree: storing balances instead of heights

* Ed Schouten <> [2015-12-10 14:12:08 +0100]:
> 2015-12-10 13:14 GMT+01:00 Szabolcs Nagy <>:
> > performance also depends on the allocator since insert/delete
> > has to malloc/free (yet another issue with the api) musl's
> > allocator is i think still better for realtime systems than
> > the jemalloc used in cloudlibc, but it will have worse average
> > performance in benchmarks like this.
> Yes. All of the tests were run on Linux, using glibc. Only the
> tsearch()/tdelete() implementations are different between tests. They
> all use glibc's standard malloc().

ah ok.

based on the updated stats the iterative bottom up
approach seems to be a good tradeoff, i wouldn't
try to reduce stack usage further by increasing the
code size.

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.