[DATAPOINT] -ac16+ and increased readahead (ksoftirqd related?)

From: Roger Larsson (roger.larsson@skelleftea.mail.telia.com)
Date: Fri Sep 28 2001 - 20:08:15 EST


After receiving Robert Cohens mail concerning
"[BENCH] Problems with IO throughput and fairness with 2.4.10 and 2.4.9-ac15"
I replied about the possibility to increase READAHEAD again.

But since I was running the -ac16 (with swapoff patch) I decided to try it

First my attempts were not that sucessful. Changing /proc/sys/vm/max-readahead
did not have any visible effect... (why not???)

Then I found /proc/ide/hda/settings
and changed "file_readahead:511" but got wery strange results...

The start is very good but suddenly the HD starts to sound strange. A check
with vmstat shows that there is a huge amount of context switches...

Going back to zero (global) readahead did not help.
So I rebooted the computer for more careful testing.

vmstat-3 is output from vmstat 3
file_readahead.settings is how I change the file_readahead

The test I perform, for each value of file_readahead, is a diff between
two huge, but identical, files.


Roger Larsson

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Sun Sep 30 2001 - 21:01:07 EST