public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH] readahead: fix initial window size calculation
@ 2006-02-26 19:50 Oleg Nesterov
  2006-02-27 15:41 ` Steven Pratt
  0 siblings, 1 reply; 2+ messages in thread
From: Oleg Nesterov @ 2006-02-26 19:50 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Bryan Fink, linux-kernel, Ram Pai, Steven Pratt, Trond Myklebust

(I hope it is ok to send this patch without Steven's reply)

From: Steven Pratt <slpratt@austin.ibm.com>

The current current get_init_ra_size is not optimal across different IO
sizes and max_readahead values.  Here is a quick summary of sizes
computed under current design and under the attached patch.  All of
these assume 1st IO at offset 0, or 1st detected sequential IO.

32k max, 4k request

old         new
-----------------
 8k        8k
16k       16k
32k       32k

128k max, 4k request
old         new
-----------------
32k         16k
64k         32k
128k        64k
128k       128k

128k max, 32k request
old         new
-----------------
32k         64k    <-----
64k        128k
128k       128k

512k max, 4k request
old         new
-----------------
4k         32k     <----
16k        64k
64k       128k
128k      256k
512k      512k

--- 2.6.16-rc3/mm/readahead.c~	2006-02-27 00:53:17.881019192 +0300
+++ 2.6.16-rc3/mm/readahead.c	2006-02-27 01:10:39.172718792 +0300
@@ -83,10 +83,10 @@ static unsigned long get_init_ra_size(un
 {
 	unsigned long newsize = roundup_pow_of_two(size);
 
-	if (newsize <= max / 64)
-		newsize = newsize * newsize;
+	if (newsize <= max / 32)
+		newsize = newsize * 4;
 	else if (newsize <= max / 4)
-		newsize = max / 4;
+		newsize = newsize * 2;
 	else
 		newsize = max;
 	return newsize;

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2006-02-27 15:43 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-02-26 19:50 [PATCH] readahead: fix initial window size calculation Oleg Nesterov
2006-02-27 15:41 ` Steven Pratt

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox