Re: [9fans] Re: Thinknic and plan9

Re: [9fans] Re: Thinknic and plan9

[presotto]

[-- Attachment #1: Type: text/plain, Size: 86 bytes --]

Is BSD trying to reverse lookup the address of the Plan 9 box or something
similar?

[-- Attachment #2: Type: message/rfc822, Size: 4740 bytes --]

From: Dan Cross <cross@math.psu.edu>
To: 9fans@cse.psu.edu
Cc: 
Subject: [9fans] Re: Thinknic and plan9
Date: Thu, 31 May 2001 12:30:04 -0400 (EDT)
Message-ID: <200105311630.MAA23448@augusta.math.psu.edu>

In article <slrn9h01d5.in7.randolph@panix3.panix.com> you write:
>I ran across this inexpensive little diskless (it has a CD-ROM) x86
>box at www.thinknic.com and I think might make an excellent Plan 9
>terminal.  Has anyone tried it?

Hmm, I started looking into this more, because such things interest
me.  The thing is certainly cheap ($199, plus shipping and tax).
Unfortunately, most of the hardware is unsupported.  Well, okay, just
about everything is unsupported.

The ethernet chip is, I believe, an SiS900.  A Plan 9 driver doesn't
exist for this specific chip; I don't know if any of the existing
drivers would run it.  However, a data sheet on the chip is available
from the SiS web site, and there are Linux and BSD drivers from which
to build.  I'll see if I can't get something going.

The video chip is an SiS5597/98.  I know nothing about it, except that
apparantly it only has 512K of RAM on it.  XFree86 seems to have a
driver for it (complete with acceleration and the whole bit), but
that's about all I know.  The 512KB of RAM is a little confusing; it
doesn't seem like enough to run it at the resolution and color depth
stated on the thinknic web page; maybe it steals RAM from the host.  I
don't know.  I couldn't find a data sheet on it, but I didn't look that
hard.

The audio chip is a C-Media device (CMI8738) which also drives the
modem.  I can't imagine that it's supported, but again, a Linux driver
exists.  I haven't even attempted to look for a data sheet.

Other devices in the box are an SiS 7001 USB controller (unsupported),
and an SiS 85C5513 IDE controller.  I don't know much about the IDE
controller, or its software interface, so I can't say wether or not
it's supported.  The device also has a CD-ROM and a 4MB IDE flash
disk.  Yes, that's four megabytes; I gather they only support using it
to hold things like netscape bookmarks, etc.  It'd certainly be big
enough to hold a Plan 9 kernel and bootstrap.

I gather that the thing uses the CD-ROM for booting and housing the
operating system (Linux).  Booting off of the flash would free that up,
and make it useful for other stuff.  If one could access the flash as a
normal, if somewhat small, IDE device, and the IDE controller is
supported, then that more or less takes care of that problem.

Apparantly, some people have had success with putting laptop hard
drives into these units, but it involves hardware hacking.

Since they're so cheap, I decided to ship eatting out for the next few
days and order one.  It should be here on the 6th of June.  In the mean
time, I'll try and build device drivers for the ethernet and video
chips; maybe I'll have something ready for testing by the time the unit
gets here.

btw- I think that the addition of yet another SB-incompatable audio
chip to the mix underscores the need to divorce the kernel and external
audio interfaces from the soundblaster family.  Having a modularized
framework for audio devices, much the way the current ethernet
interface has been modularized, and moving towards a standard such as
PCM for transfer of audio data into and out of the kernel, would be a
good step towards doing that.

	- Dan C.

Re: [9fans] Re: Thinknic and plan9

[Dan Cross]

In article <20010531204829.DF5B1199E9@mail.cse.psu.edu> you write:
>i was going by the specs you published (512KB video memory - that won't
>even get you to 1024x768x8), but it's clearly better than that from the
>little info on the web page.

Yeah, I just cribbed that data from several web pages; the 512KB spec
can't be right.

>i've nothing against it in principle, i'd love to replace this hot and
>noisy pc under my desk with something that does only the things i need -
>100base-t and a display. but if the display is limited to 1024x768
>or less then i'm better off using my laptop.

Yeah, I agree.  I'll see what I can get cooking.  At least the ethernet
chip is relatively simple.

	- Dan C.

Re: [9fans] Re: Thinknic and plan9

[Russ Cox]

i think the 512kb video memory is a lie.
it sounds like the sis all-in-one board
that leeches video memory from the system memory.

russ

Re: [9fans] Re: Thinknic and plan9

[jmk]

On Thu May 31 15:58:06 EDT 2001, cross@math.psu.edu wrote:
> In article <20010531171819.07444199C0@mail.cse.psu.edu> you write:
> >Unsupported by current software can be fixed.
> 
> Yes.
> 
> >Inadequate hardware that's fixed (the video hardware) can't.
> 
> The Linux people don't seem to be having a problem using the inadequate
> video hardware at high resolution and with many colors.

i was going by the specs you published (512KB video memory - that won't
even get you to 1024x768x8), but it's clearly better than that from the
little info on the web page.

i've nothing against it in principle, i'd love to replace this hot and
noisy pc under my desk with something that does only the things i need -
100base-t and a display. but if the display is limited to 1024x768
or less then i'm better off using my laptop.

Re: [9fans] Re: Thinknic and plan9

[Dan Cross]

In article <20010531194342.B7DD519A33@mail.cse.psu.edu> you write:
>As a bit of amusement, Oracle was very close (back in fairly
>early Lucent Inferno BU time) to licensing Inferno for
>this or similar boxes.  It was close enough that Ellison
>and Rich McGinn had dinner to shake hands over the deal.
>Then, quite suddenly, the deal fell apart, preumably
>not because the dinner went poorly.

McGinn questioned Ellison's divinity?

	- Dan C.

Re: [9fans] Re: Thinknic and plan9

[Fariborz 'Skip' Tavakkolian]

It is surprising to know that anyone could hold dinner
down in presence of Ellison.

At 03:43 PM 5/31/01 -0400, you wrote:
> > Oracle have been trying to market this box under various
> > guises for something like 5 years, they should give up.
>
>As a bit of amusement, Oracle was very close (back in fairly
>early Lucent Inferno BU time) to licensing Inferno for
>this or similar boxes.  It was close enough that Ellison
>and Rich McGinn had dinner to shake hands over the deal.
>Then, quite suddenly, the deal fell apart, preumably
>not because the dinner went poorly.
>
>	Dennis
>
>

Re: [9fans] Re: Thinknic and plan9

[Dan Cross]

In article <20010531171819.07444199C0@mail.cse.psu.edu> you write:
>Unsupported by current software can be fixed.

Yes.

>Inadequate hardware that's fixed (the video hardware) can't.

The Linux people don't seem to be having a problem using the inadequate
video hardware at high resolution and with many colors.

(Of course, they probably have 80 billion lines of code to support it,
but that's another matter all together.)

>Oracle have been trying to market this box under various
>guises for something like 5 years, they should give up.

Yeah, so should those pesky timesharing people.  Batch mode computing
is the only thing which is efficient enough to make effective use of
all computers, which are big, expensive centralized things with
multi-million dollar pricetags.  Oh!  Wait a minute....wrong century.
:-)

The reason I think Oracle hasn't been very successful to date is simply
marketing; particularly the Microsoft fueled belief on the part of
consumers that they ``need'' a PC.  They don't.  My Mom certainly
doesn't, and she's the type of user this thing is targetted to.

	- Dan ``It's the Marketing, Stupid!'' C.

Re: [9fans] Re: Thinknic and plan9

[dmr]

 > Oracle have been trying to market this box under various
 > guises for something like 5 years, they should give up.

As a bit of amusement, Oracle was very close (back in fairly
early Lucent Inferno BU time) to licensing Inferno for
this or similar boxes.  It was close enough that Ellison
and Rich McGinn had dinner to shake hands over the deal.
Then, quite suddenly, the deal fell apart, preumably
not because the dinner went poorly.

	Dennis

Re: [9fans] Re: Thinknic and plan9

[jmk]

On Thu May 31 12:31:34 EDT 2001, cross@math.psu.edu wrote:
> In article <slrn9h01d5.in7.randolph@panix3.panix.com> you write:
> >I ran across this inexpensive little diskless (it has a CD-ROM) x86
> >box at www.thinknic.com and I think might make an excellent Plan 9
> >terminal.  Has anyone tried it?
> 
> Hmm, I started looking into this more, because such things interest
> me.  The thing is certainly cheap ($199, plus shipping and tax).
> Unfortunately, most of the hardware is unsupported.  Well, okay, just
> about everything is unsupported.

Unsupported by current software can be fixed.
Inadequate hardware that's fixed (the video hardware) can't.
Oracle have been trying to market this box under various
guises for something like 5 years, they should give up.

[9fans] Re: Thinknic and plan9

[Dan Cross]

In article <slrn9h01d5.in7.randolph@panix3.panix.com> you write:
>I ran across this inexpensive little diskless (it has a CD-ROM) x86
>box at www.thinknic.com and I think might make an excellent Plan 9
>terminal.  Has anyone tried it?

Hmm, I started looking into this more, because such things interest
me.  The thing is certainly cheap ($199, plus shipping and tax).
Unfortunately, most of the hardware is unsupported.  Well, okay, just
about everything is unsupported.

The ethernet chip is, I believe, an SiS900.  A Plan 9 driver doesn't
exist for this specific chip; I don't know if any of the existing
drivers would run it.  However, a data sheet on the chip is available
from the SiS web site, and there are Linux and BSD drivers from which
to build.  I'll see if I can't get something going.

The video chip is an SiS5597/98.  I know nothing about it, except that
apparantly it only has 512K of RAM on it.  XFree86 seems to have a
driver for it (complete with acceleration and the whole bit), but
that's about all I know.  The 512KB of RAM is a little confusing; it
doesn't seem like enough to run it at the resolution and color depth
stated on the thinknic web page; maybe it steals RAM from the host.  I
don't know.  I couldn't find a data sheet on it, but I didn't look that
hard.

The audio chip is a C-Media device (CMI8738) which also drives the
modem.  I can't imagine that it's supported, but again, a Linux driver
exists.  I haven't even attempted to look for a data sheet.

Other devices in the box are an SiS 7001 USB controller (unsupported),
and an SiS 85C5513 IDE controller.  I don't know much about the IDE
controller, or its software interface, so I can't say wether or not
it's supported.  The device also has a CD-ROM and a 4MB IDE flash
disk.  Yes, that's four megabytes; I gather they only support using it
to hold things like netscape bookmarks, etc.  It'd certainly be big
enough to hold a Plan 9 kernel and bootstrap.

I gather that the thing uses the CD-ROM for booting and housing the
operating system (Linux).  Booting off of the flash would free that up,
and make it useful for other stuff.  If one could access the flash as a
normal, if somewhat small, IDE device, and the IDE controller is
supported, then that more or less takes care of that problem.

Apparantly, some people have had success with putting laptop hard
drives into these units, but it involves hardware hacking.

Since they're so cheap, I decided to ship eatting out for the next few
days and order one.  It should be here on the 6th of June.  In the mean
time, I'll try and build device drivers for the ethernet and video
chips; maybe I'll have something ready for testing by the time the unit
gets here.

btw- I think that the addition of yet another SB-incompatable audio
chip to the mix underscores the need to divorce the kernel and external
audio interfaces from the soundblaster family.  Having a modularized
framework for audio devices, much the way the current ethernet
interface has been modularized, and moving towards a standard such as
PCM for transfer of audio data into and out of the kernel, would be a
good step towards doing that.

	- Dan C.