[linux-audio-dev] RE: linux-audio-dev digest, Vol 1 #127 - 2 msgs

[Brad Arant]

Havent been receiving any news lately, please help.

-----Original Message-----
From: linux-audio-dev-admin at music.columbia.edu
[mailto:linux-audio-dev-admin at music.columbia.edu]On Behalf Of
linux-audio-dev-request at music.columbia.edu
Sent: Saturday, July 27, 2002 9:00 AM
To: linux-audio-dev at music.columbia.edu
Subject: linux-audio-dev digest, Vol 1 #127 - 2 msgs


Send linux-audio-dev mailing list submissions to
	linux-audio-dev at music.columbia.edu

To subscribe or unsubscribe via the World Wide Web, visit
	http://music.columbia.edu/mailman/listinfo/linux-audio-dev
or, via email, send a message with subject or body 'help' to
	linux-audio-dev-request at music.columbia.edu

You can reach the person managing the list at
	linux-audio-dev-admin at music.columbia.edu

When replying, please edit your Subject line so it is more specific
than "Re: Contents of linux-audio-dev digest..."


Today's Topics:

   1. Re: runtime / compiletime processing proof of concept (Paul Winkler)
   2. Re: runtime / compiletime processing proof of concept (Juan Linietsky)

--__--__--

Message: 1
Date: Fri, 26 Jul 2002 13:00:10 -0400
From: Paul Winkler <pw_lists at slinkp.com>
To: linux-audio-dev at music.columbia.edu
Subject: Re: [linux-audio-dev] runtime / compiletime processing proof of
concept
Reply-To: linux-audio-dev at music.columbia.edu

On Fri, Jul 26, 2002 at 01:21:46PM +0200, Maarten de Boer wrote:
> Hello,
>
> Juan Linietsky (reduz on #lad) talked on #lad about an idea of
> Benno to have processing graphs generate source code, compile
> it on the fly, thus taking advantage of all compiled optimizations,
> as an alternative to runtime processing. Especially for sample
> based processing instead of block based processing this can be
> very useful and a lot faster.

John Lazarro tells me that sfront does something similar.
He says it's described here:
http://www.cs.berkeley.edu/~lazzaro/sa/pubs/pdf/wemp01.pdf

--PW


--__--__--

Message: 2
Date: Fri, 26 Jul 2002 14:38:40 -0300
From: Juan Linietsky <coding at reduz.com.ar>
To: linux-audio-dev at music.columbia.edu
Subject: Re: [linux-audio-dev] runtime / compiletime processing proof of
concept
Organization: Codenix
Reply-To: linux-audio-dev at music.columbia.edu

On Fri, 26 Jul 2002 13:56:55 +0200
Maarten de Boer <mdeboer at iua.upf.es> wrote:

>
> What i forgot to mention:
>
> The most interesting thing is that the code used for runtime, and
> the generated code used for on the fly compilation is actually the
> same! Which ensures that the result of runtime execution and
> compiling generated code -> execution is the same, and everything
> has to be written only once.
>
> maarten
>


Hi! Thanks for taking the time of writing a proof of concept program.
So far from what i've been reading it seems ok (didnt have much time
to check the entire code in detail) althought the real
advantage of compiling such graphs is when they are very large and
stack/buffer and indirection operations start to really slow it down.
Still, even whith this, i get (please tell me if these are correct):

realtime code: (./main)

real    0m1.005s
user    0m0.970s
sys 0m0.030s

runtime code: (./gen)

real    0m0.435s
user    0m0.370s
sys 0m0.060s


three seems to be quite a difference!

Juan Linietsky





--__--__--

_______________________________________________________
linux-audio-dev Mailing List Digest
http://music.columbia.edu/mailman/listinfo/linux-audio-dev

End of linux-audio-dev Digest