Subject: Re: 1.5.1 SPL0 fix??
To: Simon Burge <>
From: Eric Schnoebelen <>
List: port-alpha
Date: 04/14/2001 12:12:09
Simon Burge writes:
- Eric Schnoebelen wrote:
- > Eric Schnoebelen writes:
- > - Can I get a patch for it, to see if it clears up the
- > - disk/networking performance problem I'm seeing on my PC164?
- > 
- > 	Well, having applied the patch, I can say that it
- > doesn't help the network/disk-io problem I've seen and others
- > have reported.
- > 
- > 	Problem summary:
- > 		With simlutainous disk and network IO, the
- > 		network IO is about 6kbp/sec.
- > 
- > 	System Description:
- > 		AlphaPC164/500Mhz
- > 		64Mb RAM
- > 		IDE disk
- > 		Intel Etherexpress Pro/100+ (fxp)
- > 		NetBSD 1.5.1_BETA kernel
- So if you're scp or ftp'ing a file from local disk (for example) you're
- saying network IO sucks?

	Yup..  I'm seeing 15KB/s from ftp on a 374Mb file..

	I'm seeing 6KB/sec from an rdump to /dev/null on
another machine on same segment.. (the network is 10baseTX-fdx.)

-                          I don't see this on my 1.5 PC164 w/epic or my
- 1.5U PC164 w/fxp boxes.  The 1.5 box is using raidframe mirrors with one
- disk on each IDE channel, the 1.5U box has a single IDE disk.  I've only
- got a 10Mbit hub, but I regularly see around a 1MByte/sec transferring
- files between the two.

	I suppose it's possible I've got a buggered mainboard,
but the system seems fine otherwise..  

	I have tried various tulip boards (using both tlp and
de), an rtk board, and the fxp currently in there.. Little
change between the various network adapters. (sadly.)

	Thank you for any and all assistance,

Eric Schnoebelen
      Another day done; All targets met; All systems fully operational;
	All customers satisfied; All staff keen and well adjusted;
		  All pigs fed and ready for take off;