| |
PA2AGA > TCPDIG 30.08.96 03:49l 179 Lines 7105 Bytes #-10903 (0) @ EU
BID : TCP_96_174A
Read: GUEST
Subj: TCP-Group Digest 96/174A
Path: DB0AAB<DB0PV<DB0WGS<OE3XSR<OK0PBX<OK0PKR<OK0PBB<OK0PAB<OH3RBR<OH2RBI<
OH7RBA<PI8VNW
Sent: 960829/2002Z @:PI8VNW.#ZH2.NLD.EU #:57793 [Hoek v Holland] FBB5.15c
From: PA2AGA@PI8VNW.#ZH2.NLD.EU
To : TCPDIG@EU
Received: from pa2aga by pi1hvh with SMTP
id AA16418 ; Thu, 29 Aug 96 19:37:42 UTC
Received: from pa2aga-1 by pa2aga (NET/Mac 2.3.61/7.1) with SMTP
id AA00003645 ; Thu, 29 Aug 96 20:54:53 MET
Received: from pa2aga-1 by pa2aga-1 (NET/Mac 2.3.61/7.5.4) with SMTP
id AA00007983 ; Thu, 29 Aug 96 20:39:58 MET
Date: Thu, 29 Aug 96 20:29:47 MET
Message-Id: <tcp_96_174A>
From: pa2aga
To: tcp_broadcast@pa2aga-1
Subject: TCP-Group Digest 96/174A
X-BBS-Msg-Type: B
TCP-Group Digest Wed, 28 Aug 96 Volume 96 : Issue 174
Today's Topics:
advanced networking (5 msgs)
AX.25 - that darn PID (2 msgs)
Linux router boot floppy (2 msgs)
Looking for an RFC (2 msgs)
PPP/SLIP in connected AX25? (18 msgs)
your mail
Send Replies or notes for publication to: <TCP-Group@UCSD.Edu>.
Subscription requests to <TCP-Group-REQUEST@UCSD.Edu>.
Problems you can't solve otherwise to brian@ucsd.edu.
Archives of past issues of the TCP-Group Digest are available
(by FTP only) from ftp.UCSD.Edu in directory "mailarchives".
We trust that readers are intelligent enough to realize that all text
herein consists of personal comments and does not represent the official
policies or positions of any party. Your mileage may vary. So there.
----------------------------------------------------------------------
Date: Tue, 27 Aug 1996 10:03:24 +0100 (BST)
From: Alan Cox <alan@cymru.net>
Subject: advanced networking
Most of the low end intel hardware isnt a good economic move simply because
the difference between the average modern 386sx board and the big 286 boards
in power consumption is going to bite you in about a year
There are some very nice embedded NS 486 chips now with serial, bus
interfaces and all sorts if you wanted to go that way. It would need a bit
of work as it has no real mode and no paging (just segments).
Alan
------------------------------
Date: Tue, 27 Aug 1996 09:33:42 -0500
From: "Steven R. Sampson" <ssampson@othello.tinker.af.mil>
Subject: advanced networking
Alan Cox wrote:
>
> There are some very nice embedded NS 486 chips now with serial, bus
> interfaces and all sorts if you wanted to go that way.
One chip that I was initially impressed with, was the 486 SLC chip
by IBM. I purchased a mother board with this chip, and a seperate
math coprocessor. The SLC chip is a little bitty thing. You would
expect the performance to be less than a real 486, but if bogomips
are any indication, it is a real dog, and a candidate for a router.
I get 19 bogomips on a 66 MHz version with redhat linux. The same
linux runs 36 bogomips on a 486 DX2 66 MHz board. The other problem
is the GNU C compiler. When I compile the kernel I get a segment
violation (probably a timeout). What this means, is that I go back to
the last file it was trying to compile, delete the .o file, and then
type make again. It takes about 8 of these blurps to get through the
complete compile. On the DX2 chip it compiles straight through.
I've had this SLC chip for over a year. It worked just fine with WinWG
3.11, but when I switched to Win95 it caved in gasping for air.
Steve
------------------------------
Date: Tue, 27 Aug 1996 17:44:56 +0100 (BST)
From: Alan Cox <alan@cymru.net>
Subject: advanced networking
> math coprocessor. The SLC chip is a little bitty thing. You would
> expect the performance to be less than a real 486, but if bogomips
> are any indication, it is a real dog, and a candidate for a router.
Be very careful. The bogomips rating in Linux is bogo - for bogus, its
a tight timing loop. The NS chips are also a bit different in that
they have a lot more on board I/O and really are targetted at embeedding
> linux runs 36 bogomips on a 486 DX2 66 MHz board. The other problem
> is the GNU C compiler. When I compile the kernel I get a segment
> violation (probably a timeout). What this means, is that I go back to
> the last file it was trying to compile, delete the .o file, and then
> type make again. It takes about 8 of these blurps to get through the
> complete compile. On the DX2 chip it compiles straight through.
Your SLC chip is faulty or overheating, or you've got cache/bus problems
if its doing that. I had to slap a CPU fan on a genuine 486DX2/66 to
stop my box misbehaving[1].
Alan
[1] yes I know the 486DX2/66 5v is meant to have a fan, I guess the
vendor didnt believe intel on that.
------------------------------
Date: Wed, 28 Aug 1996 08:56:17 +1000
From: Terry Dawson <terry@perf.no.itg.telstra.com.au>
Subject: advanced networking
> > it is really quite trivial. It is just a combination of the kernel
> > supported compressed root filesystem, ram disk and a bare bones root
> > filesystem containing sufficient tools to build a device.
>
> Note that the 'yard' utility available in sunsite's Linux/Incomming
> directory makes it very easy to build a custom boot boot floppy
> with your own kernel and a ramdisk image. Perhaps you could describe
> the set of files needed for Bootdisk_Contents and the startup
> scripts.
I haven't looked at 'yard' yet, though I've been meaning to. At present
I've just used a shell script to do the work. I'll take a look, and if
the
yard solution is more elegant than mine I'll do as you suggest.
Terry
------------------------------
Date: Wed, 28 Aug 96 02:43:00 -0000
From: mikebw@bilow.bilow.uu.ids.net (Mike Bilow)
Subject: advanced networking
Steven R. Sampson wrote in a message to Mike Bilow:
SRS> One chip that I was initially impressed with, was the 486
SRS> SLC chip by IBM. I purchased a mother board with this chip,
SRS> and a seperate math coprocessor. The SLC chip is a little
SRS> bitty thing. You would expect the performance to be less
SRS> than a real 486, but if bogomips are any indication, it is a
SRS> real dog, and a candidate for a router.
Despite the number, the CPU core is a 386. However, it does outperform a real
386 running at the same clock speed.
SRS> I get 19 bogomips on a 66 MHz version with redhat linux.
SRS> The same linux runs 36 bogomips on a 486 DX2 66 MHz board.
SRS> The other problem is the GNU C compiler. When I compile the
SRS> kernel I get a segment violation (probably a timeout). What
SRS> this means, is that I go back to the last file it was trying
SRS> to compile, delete the .o file, and then type make again.
SRS> It takes about 8 of these blurps to get through the complete
SRS> compile. On the DX2 chip it compiles straight through.
Failures of this kind are usually a result of deficiencies in the motherboard,
especially in the L2 cache. You may also have to enable the "OS/2 Compatible"
setting in the CMOS setup if you intend to run the chip in protected mode.
The
real IBM motherboards ("Blue Lightning") offer this option.
SRS> I've had this SLC chip for over a year. It worked just fine
SRS> with WinWG 3.11, but when I switched to Win95 it caved in
To be continued in digest: tcp_96_174B
Read previous mail | Read next mail
| |
