lkml.org 
[lkml]   [2003]   [Oct]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: VIA IDE performance under 2.6.0-test7/8?
Hi James!

> Can you also send the output from "cat /proc/interrupts".
> It looks like you are not using IO-APIC, but instead using XT-PIC.
> XT-PIC is a lot slower than IO-APIC.

Reading the previous messages I compared the SuSE 2.4.20-4GB and the
2.6.0-test8-bk1 kernel.

The mainboard is a VIA EPIA 5000 (cpu Via Eden 533 Mhz, Via 8601A
Northbridge, VT8231 Southbridge)

hdparm -I /dev/hd? displays no difference ... udma2 is the used mode.

hdparm -T /dev/hd? measures transfer rates of 60-63 MB/s, there is no
significant difference between
the kernel versions.

hdparm -t /dev/hd? mesasures 18 MB/s for kernel 2.6.0-test8-bk1 and 27
MB/s for kernel 2.4.20.
This is a significant :-(

2.4.20 gives almost exactly 50% better performance compared to
2.6.0-test8 ... is this pure accident or could this give a hint?

> Just turn on SMB support in the "make menuconf", and it should enable
> IO-APIC.

Compiling the kernel with and without smb support as well as trying the
other APIC related new configuration options
does change nothing. There is allways a "No local APIC present or
hardware disabled" message. I believe that there
really is no IO-APIC, at least I found no related BIOS configuration option.

/proc/interrupts indicates an XT-PIC for both kernel versions.

IRQ setup is identical, IDE IRQs are not shared with any other devices.

The drive is the only drive attached.

Playing around with different read-ahead values does not help.

Any ideas?

cu,
Knut Petersen


-
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/

\
 
 \ /
  Last update: 2005-03-22 13:58    [W:0.049 / U:1.924 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site