From: "Robert G. Brown" <rgb@phy.duke.edu>
To: Rhys Ulerich <rhys.ulerich@gmail.com>
Cc: gsl-discuss@sourceware.org
Subject: Re: gsl container designs
Date: Thu, 07 Jan 2010 13:22:00 -0000 [thread overview]
Message-ID: <alpine.LFD.2.00.1001070815530.2256@localhost> (raw)
In-Reply-To: <4a00655d1001062146g555fd9dfh1e333613d8e3b463@mail.gmail.com>
[-- Attachment #1: Type: TEXT/PLAIN, Size: 1760 bytes --]
On Wed, 6 Jan 2010, Rhys Ulerich wrote:
>> I recall from my benchmarking days that -- depending on compiler --
>> there is a small dereferencing penalty for packed matrices (vectors
>> packed into dereferencing **..* pointers) compared to doing the offset
>> arithmetic via brute force inline or via a macro.
>> ......
>> I haven't
>> run the benchmark recently and don't know how large it currently is. Â It
>> was never so large that it stopped me from using repacked pointers for
>> code clarity..
>
> Mostly unscientific, but worth tossing into the mix:
>
> Using Intel 10.1 compilers on a fairly recent AMD chip, 100,000 iterations
> of doing the nested pointers approach is neck-and-neck with index arithmetic
> on a 10x10 double matrix. Â For the 100x100 case it takes 1.3 times longer
> to iterate using the nested pointers. Â Work in the inner loop "compute
> kernel" is
> *= against a constant scalar. Â Optimization flags on -O3. Â I've seen similar
> behavior on recent GNU compilers.
That sounds partly like a cache effect -- 10x10 almost certainly stays
in L1, 100x100 won't fit. My own experience is similar, although I
don't recall the multiplier being as large as 1.3 (but then, I was doing
stream and stream-like tests with very large vectors, which means that
one spends more time in a vector streaming mode and minimizes
cache-thrashing when turning corners). And my memory could be faulty --
I'm an old guy, after all, early Alzheimers...;-)
rgb
>
> I'm happy to provide the test code if anyone's interested.
>
> - Rhys
>
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu
next prev parent reply other threads:[~2010-01-07 13:22 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-25 0:55 Gerard Jungman
2009-11-29 21:04 ` Brian Gough
2009-12-04 18:48 ` Brian Gough
2009-12-09 21:04 ` using GSL with C++ (was Re: gsl container designs) James Amundson
2009-12-09 21:14 ` Jochen Küpper
2009-12-09 21:54 ` Jari Häkkinen
2009-12-10 11:46 ` Brian Gough
2009-12-10 12:09 ` Brian Gough
2009-12-10 13:42 ` Robert G. Brown
2009-12-10 21:44 ` James Bergstra
2009-12-10 15:15 ` Kevin H. Hobbs
2010-01-06 11:45 ` gsl container designs Tuomo Keskitalo
2010-01-06 15:47 ` Robert G. Brown
2010-01-07 1:50 ` Gerard Jungman
2010-01-07 3:29 ` Robert G. Brown
[not found] ` <4a00655d1001062110m139c0a8tf2eae7de67da8f6f@mail.gmail.com>
2010-01-07 5:46 ` Rhys Ulerich
2010-01-07 13:22 ` Robert G. Brown [this message]
2010-01-08 20:08 ` Gerard Jungman
2010-01-07 18:29 ` Brian Gough
2010-01-06 12:04 ` Tuomo Keskitalo
2010-01-06 19:57 ` Gerard Jungman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=alpine.LFD.2.00.1001070815530.2256@localhost \
--to=rgb@phy.duke.edu \
--cc=gsl-discuss@sourceware.org \
--cc=rhys.ulerich@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).