[MLton-devel] forwarded message from Henry Cejtin

[Stephen Weeks]

--TQwEh0cUQ9
Content-Type: text/plain; charset=us-ascii
Content-Description: message body text
Content-Transfer-Encoding: 7bit


Hey guys, let's try to keep MLton@mlton.org and
andrew.wright@aleri.com on this thread.  Thanks.


--TQwEh0cUQ9
Content-Type: message/rfc822
Content-Description: forwarded message
Content-Transfer-Encoding: 7bit

Return-path: <henry@sourcelight.com>
Received: from pop3-sjc.bbnow.net [24.219.4.120]
	by localhost with POP3 (fetchmail-5.7.4)
	for sweeks@localhost (single-drop); Thu, 16 May 2002 11:30:05 -0700 (PDT)
Received: from i3s.net (unverified [24.219.4.129]) by i3s.net
 (Rockliffe SMTPRA 4.5.4) with ESMTP id <B0003035718@i3s.net> for <sweeks@bbnow.net>;
 Thu, 16 May 2002 13:26:10 -0500
Received: from mail.san.yahoo.com (unverified [209.132.1.30]) by i3s.net
 (Rockliffe SMTPRA 4.5.4) with ESMTP id <B0001336069@i3s.net> for <sweeks@bbnow.net>;
 Thu, 16 May 2002 13:26:10 -0500
Received: from syzygy.sourcelight.com (199.249.165.245) by mail.san.yahoo.com (6.5.017.1)
        id 3CE3917C00021C59 for sweeks@bbnow.net; Thu, 16 May 2002 11:25:33 -0700
Received: (from henry@localhost)
	by syzygy.sourcelight.com (8.11.2/8.11.2) id g4GIPhg19943;
	Thu, 16 May 2002 13:25:43 -0500
Message-Id: <200205161825.g4GIPhg19943@syzygy.sourcelight.com>
From: Henry Cejtin <henry@sourcelight.com>
To: Suresh.Jagannathan@storagenetworks.com, sweeks@sweeks.com
Subject: Re: [MLton-devel] pinned objects
Date: Thu, 16 May 2002 13:25:43 -0500

I'm with Suresh, but have a stronger statement: the only place I can imagine
pinning for I/O being needed is in Windows asynchronus I/O.  Is that what
you are thinking of it for?  Why not just bite the bullet and allocate the
space in C.  In my experience, using mmap for file I/O is really only worth
while when you are doing lots of random seeks around in the file, since then
it avoids the kernel-user copy you would have to do otherwise to freshen
the cache which is user space.  This doesn't happen very often.
In fact, I think that the only time I actually used it was in the protein
folding code, and the only reason I did it there was to have the 2 processes
share the same memory used for the tree.  Definitely a hack.

--TQwEh0cUQ9--

_______________________________________________________________

Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: bandwidth@sourceforge.net
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel