Re: [9fans] Are nvidia-cards working with plan9?

Re: [9fans] Are nvidia-cards working with plan9?

[David Gordon Hogan]

> > If the Wall Street crowd had their way, Microsoft would
> > declare their drop-down menus proprietary, so you'd have
> > to hire someone to operate your PCs for you.
> >
> Don't blame Wall Street, blame the buying public, and their inability
> to resist the media.

Or blame the media for being so hard to resist...  (I sense an infinite
regress lurking here).

Re: [9fans] Are nvidia-cards working with plan9?

[jiho]

In article <20001005151052.3FEAB199DE@mail>,
	dhog@plan9.bell-labs.com (David Gordon Hogan) writes:

>>> If the Wall Street crowd had their way, Microsoft would
>>> declare their drop-down menus proprietary, so you'd have
>>> to hire someone to operate your PCs for you.
>>
>> Don't blame Wall Street, blame the buying public, and their inability
>> to resist the media.
>
> Or blame the media for being so hard to resist...  (I sense an infinite
> regress lurking here).

Actually, I wasn't laying blame, I was just alluding
to _de_facto_ reality.

There's a small matter of U.S. anti-trust law, and a
principle known as "fair access to essential
facilities".  It dates back to the railroad tycoons,
fer cryin' out loud.  Ironically, the case has been
made against Microsoft software.  It's even more
obvious with respect to chips, but as far as I know
(not very far) no one has tried making the case.

In this context "fair access" means public
documentation, sufficient for any competent graphics
chip driver writer to write his own "naked hardware"
driver -- for any arbitrary OS -- which uses all
features of the chip claimed by the chip's maker.

In other words, legally you shouldn't _need_ to burn
your own silicon, at least not in the U.S.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[Lucio De Re]

On Mon, Oct 09, 2000 at 09:05:26AM +0000, jiho@smtp.popsite.net wrote:
> Message-ID: <39e0c6c7$1@news2.starnetinc.com>
> Organization: University of Bath Computing Services, UK
> Content-Type: text/plain; charset=us-ascii
> References: <20001005151052.3FEAB199DE@mail>
> Subject: Re: [9fans] Are nvidia-cards working with plan9?
> Sender: 9fans-admin@cse.psu.edu
> Errors-To: 9fans-admin@cse.psu.edu
> X-BeenThere: 9fans@cse.psu.edu
> X-Mailman-Version: 2.0beta4
> Precedence: bulk
> Reply-To: 9fans@cse.psu.edu
> List-Id: Fans of the O/S Plan 9 from Bell Labs <9fans.cse.psu.edu>
>
> In article <20001005151052.3FEAB199DE@mail>,
>
> Actually, I wasn't laying blame, I was just alluding
> to _de_facto_ reality.
>
Well, Wall Street is just a special-interest group :-)

> There's a small matter of U.S. anti-trust law, and a
> principle known as "fair access to essential
> facilities".  It dates back to the railroad tycoons,
> fer cryin' out loud.  Ironically, the case has been
> made against Microsoft software.  It's even more
> obvious with respect to chips, but as far as I know
> (not very far) no one has tried making the case.
>
The stated reason for not pursuing the case with the full might of
the law was that _all_ IT companies had broken the law for years
(software that ran only on some hardware, applications that ran
only on given operating environment, etc) and the can of worms
would have been far too big a problem...

So we're lumped with legislation without teeth.  And the rest of
the world is hardly likely to step into the US Government shoes
and risk technological boycotts while they try to catch up :-(

++L

Re: [9fans] Are nvidia-cards working with plan9?

[jiho]

In article <20001009114925.A29182@cackle.proxima.alt.za>,
	lucio@proxima.alt.za (Lucio De Re) writes:

>> Actually, I wasn't laying blame, I was just alluding
>> to _de_facto_ reality.
>
> Well, Wall Street is just a special-interest group :-)

Yes, that certainly is a _de_facto_ reality.

>> There's a small matter of U.S. anti-trust law, and a
>> principle known as "fair access to essential
>> facilities".  It dates back to the railroad tycoons,
>> fer cryin' out loud.  Ironically, the case has been
>> made against Microsoft software.  It's even more
>> obvious with respect to chips, but as far as I know
>> (not very far) no one has tried making the case.
>
> The stated reason for not pursuing the case with the full might of
> the law was that _all_ IT companies had broken the law for years
> (software that ran only on some hardware, applications that ran
> only on given operating environment, etc) and the can of worms
> would have been far too big a problem...

Well, the stated reason is wrong.  There is a "reasonableness
clause" in the law, such that the remedy _cannot_ be unreasonably
onerous.

Requiring anyone who writes software to port and support on all
possible platforms would clearly be unreasonably onerous.

Chip companies, however, already have register-level documentation
for their own internal use, so making it public is not itself
onerous at all, especially with the Internet.  And support should
not be required, beyond keeping the documentation reasonably up
to date.

In other words, the circumstances are simply not comparable.  OS
writers shouldn't have to port to all hardware, and application
writers shouldn't have to port to all OSes.  However, OS companies
should have to provide reasonable documentation to all application
writers, and hardware complanies should have to provide reasonable
documentation to all OS writers.

> So we're lumped with legislation without teeth.  And the rest of
> the world is hardly likely to step into the US Government shoes
> and risk technological boycotts while they try to catch up :-(

It's not the legislation that lacks teeth.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 480 bytes --]

>Assuming you have (or can get) one.  Most of them are obsolete.

This may be true.
However, I can choose inferno if I really want to run such a thing
like Plan 9 on a newest hardware I have.   After I got inferno
on our Plan 9 system and on my home Windows/95 machine,
I got this feeling from deep.

Then, the reason why I'm using Plan 9 on our Univ. is very clear:
I love its clearness and simple is best principle, in other words,
I'm one of 9fans.   ^_^

Kenji


[-- Attachment #2: Type: message/rfc822, Size: 2246 bytes --]

From: jiho@smtp.popsite.net
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Are nvidia-cards working with plan9?
Date: Wed, 4 Oct 2000 08:45:25 GMT
Message-ID: <39da37e4@news2.starnetinc.com>

In article <39D90BF6.9133377@null.net>,
	"Douglas A. Gwyn" <DAGwyn@null.net> writes:

> root@mail.c-zone.net wrote:
>> For this plan to work, people must have the system installed
>> and functional.  If they need a graphics card driver first,
>> that might be difficult.
>
> It's easy enough to install a supported graphics card.

Assuming you have (or can get) one.  Most of them are obsolete.

Then there's the issue of either setting up a spare machine,
or replacing the (better) card in the machine you use....


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[Russ Cox]

	s/I'll pass along the MOT S19 records for the 1.04 bios if anyone would
	like them//

It seems to me that if replacing the BIOS is all
that you need to do to get a card working, you
should be able to just add the right BIOS string
in vgadb instead: since we don't use the BIOS,
if aux/vga works under the old BIOS it should
work under the new one.

Russ

Re: [9fans] Are nvidia-cards working with plan9?

[Nigel Roles]

You may well be right Russ. However, I thought that this
email demonstrated 'the right stuff'. Not

"changing my video card is too much trouble to run Plan 9",

but

"I'll etch my own silicon if needs be"


> 	s/I'll pass along the MOT S19 records for the 1.04 bios if anyone would
> 	like them//
>
> It seems to me that if replacing the BIOS is all
> that you need to do to get a card working, you
> should be able to just add the right BIOS string
> in vgadb instead: since we don't use the BIOS,
> if aux/vga works under the old BIOS it should
> work under the new one.
>
> Russ
>
>

Re: [9fans] Are nvidia-cards working with plan9?

[jiho]

In article <39DB3CE8.13472.13A85CF@localhost>,
	nigel@9fs.org (Nigel Roles) writes:

> You may well be right Russ. However, I thought that this
> email demonstrated 'the right stuff'. Not
>
> "changing my video card is too much trouble to run Plan 9",
>
> but
>
> "I'll etch my own silicon if needs be"

Well, there's no response to that level of fanaticism.

But there really is a serious problem with chip companies
claiming their register sets are somehow proprietary.

If the Wall Street crowd had their way, Microsoft would
declare their drop-down menus proprietary, so you'd have
to hire someone to operate your PCs for you.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[Lucio De Re]

On Thu, Oct 05, 2000 at 08:22:44AM +0000, jiho@smtp.popsite.net wrote:
>
> But there really is a serious problem with chip companies
> claiming their register sets are somehow proprietary.
>
It still baffles me why nobody has produced a graphic card that
speaks a sensible protocol instead of being variously I/O and memory
mapped in the most unorthodox manners.

For that matter, why on earth did the mouse controller migrate to
the keyboard handler, when I have yet to see a single PC clone with
a video card that did not need a mouse?

The Ontel Amigo (1979 vintage?) had dual ported memory (all of
32KB) into which the 8085 (or Z80?) would deposit a 6502 program
to drive the graphics output.  The default program was provided as
source, so you could do whatever you liked with it.

Am I being naive, or is it too late for that type of sensible
engineering to happen again?  Have Intel really disbanded the i860
team?

The other question, unfortunately, is whether there is any room
for the double Steves of the world, I mean, garage engineering
making it big?

> If the Wall Street crowd had their way, Microsoft would
> declare their drop-down menus proprietary, so you'd have
> to hire someone to operate your PCs for you.
>
Don't blame Wall Street, blame the buying public, and their inability
to resist the media.

++L

Re: [9fans] Are nvidia-cards working with plan9?

[Douglas A. Gwyn]

Lucio De Re wrote:
> It still baffles me why nobody has produced a graphic card that
> speaks a sensible protocol instead of being variously I/O and memory
> mapped in the most unorthodox manners.

A "sensible protocol" would be a flat frame buffer with
a single color depth.  Unfortunately, the vast majority of
today's customers for add-in video cards want the fastest,
most featureful 3D rendering engines.  Presumably many of
the details of the 2D interface are dictated by the needs
of the 3D design, including the bus cache.  Of course, the
"standard" VGA/SVGA modes that evolved step by step from
the original IBM VGA over the years must still be supported
for such things as Windows "Safe Mode".

> For that matter, why on earth did the mouse controller migrate
> to the keyboard handler, when I have yet to see a single PC
> clone with a video card that did not need a mouse?

The mouse was never (on the PC platform) closely coupled
with the display.  Typical PCs do not "need" a mouse, but
it is more tedious to navigate in Windows via the keyboard.

> The Ontel Amigo ... is it too late for that type of sensible
> engineering to happen again?

It really doesn't seem that such a design would be competitive
today.

> The other question, unfortunately, is whether there is any room
> for the double Steves of the world, I mean, garage engineering
> making it big?

I think you left out a "b".

Re: [9fans] Are nvidia-cards working with plan9?

[Lucio De Re]

On Mon, Oct 09, 2000 at 09:04:49AM +0000, Douglas A. Gwyn wrote:
>
> Lucio De Re wrote:
> > It still baffles me why nobody has produced a graphic card that
> > speaks a sensible protocol instead of being variously I/O and memory
> > mapped in the most unorthodox manners.
>
> A "sensible protocol" would be a flat frame buffer with
> a single color depth.  Unfortunately, the vast majority of

Surely that's not essential?  I'm thinking (but please keep in mind I'm
very much a graphics layman) of something like Tk's widgets and the means
to manipulate them as an extreme case.  3D, naturally, would require
extensions, but a language like Tcl, even if not necessarily to
everyone's tastes, would enable the construction of such extensions.  Of
course, one still needs the channel for the communication of commands and
responses between the graphic engine and the client, but is the above
idea really totally off the mark?

> > For that matter, why on earth did the mouse controller migrate
> > to the keyboard handler, when I have yet to see a single PC
> > clone with a video card that did not need a mouse?
>
> The mouse was never (on the PC platform) closely coupled
> with the display.  Typical PCs do not "need" a mouse, but
> it is more tedious to navigate in Windows via the keyboard.
>
Sorry, poor wording.  It migrated from the bus or the serial port, I
didn't mean to imply that it ever was on the display adapter.  I remember
Intel making some combination blit/mouse/network adapters in the late 80s
and early 90s, but they weren't terribly popular.  Then again, screen
capabilities in those days weren't exactly at the commodity level, and I
may be misremembering about the mouse as well.

My point is that graphics have more or less always been accompanied by
the mouse in the history of commodity computing.

> > The Ontel Amigo ... is it too late for that type of sensible
> > engineering to happen again?
>
> It really doesn't seem that such a design would be competitive
> today.
>
Considering the Tcl/Tk extreme measure mentioned above, I don't see the
obstacles.  Think Intel 860 (in today's terms) on the one side, and Power
PC on the user end.

> > The other question, unfortunately, is whether there is any room
> > for the double Steves of the world, I mean, garage engineering
> > making it big?
>
> I think you left out a "b".

Oops, I must more stupid than normally accounted for :-)  Private
explanation, please!  :-)  :-)  :-)

++L

Re: [9fans] Are nvidia-cards working with plan9?

[Rick Hohensee]

> >
> Considering the Tcl/Tk extreme measure mentioned above, I don't see the
> obstacles.  Think Intel 860 (in today's terms) on the one side, and Power
> PC on the user end.
>
> > > The other question, unfortunately, is whether there is any room
> > > for the double Steves of the world, I mean, garage engineering
> > > making it big?


Chuck Moore, author of Forth, designs CPUs in his kitchen. Not FPGAs,
fast small silicon. It takes some funding to get a fab run, but if you
have a chip, funding shouldn't be too too tough. Chuck's stuff is
weird, but it screams. So the means to make it exist. How big is another
matter. I like to think Microsoft, for example, is an unfortunate
(for the rest of us) fluke.


Rick Hohensee
r@cLIeNUX.com



> >
> > I think you left out a "b".
>
> Oops, I must more stupid than normally accounted for :-)  Private
> explanation, please!  :-)  :-)  :-)
>
> ++L
>

Re: [9fans] Are nvidia-cards working with plan9?

[Matt]

> > > The other question, unfortunately, is whether there is any room
> > > for the double Steves of the world, I mean, garage engineering
> > > making it big?
> >
> > I think you left out a "b".
>
> Oops, I must more stupid than normally accounted for :-)  Private
> explanation, please!  :-)  :-)  :-)


one presumes "garage engineering" == "garbage engineering"
and hence the macintrash was born

Re: [9fans] Are nvidia-cards working with plan9?

[forsyth]

>>I can't help but wonder where a lot of other OSes would have been
>>right now if they had not turned their noses up at 'gaming hardware'.
>>It's a good thing that 

it isn't a case of turning up noses at games or gaming hardware;
it's often a matter of manufacturers not
releasing the data required to program them, or not without one or more of:
a fuss, NDA, no source distribution, internet lobbying, manufacturer
jumping on linux bandwagon, and reverse engineering.
it's a lot of fuss for a research group once they've got drivers for
a reasonable selection of reasonable cards from reasonable manufacturers,
updated periodically as cards come and go.

Re: [9fans] Are nvidia-cards working with plan9?

[Conor]

That is all very true and I somehow don't think P9 is envisioned as a gaming
platform anytime in the near future;however,I hardly think,going by the
hardware computability list,' reasonable' comes to mind when looking at the
supported cards.

Is there something wrong with this newsgroup for it seems to take 12/24
hours to post to it?

Re: [9fans] Are nvidia-cards working with plan9?

[Douglas A. Gwyn]

Conor wrote:
> ... I hardly think,going by the hardware computability list,
> ' reasonable' comes to mind when looking at the supported cards.

By all means develop drivers for other cards that interest you,
and contribute them to the cause.  If they work at all I suspect
they'll be added to the distribution.

Re: [9fans] Are nvidia-cards working with plan9?

[root]

In article <8qnit2$37p$1@supernews.com>,
	Conor <conoruser@lineone.net> writes:

> That is all very true and I somehow don't think P9 is envisioned as a gaming
> platform anytime in the near future;however,I hardly think,going by the
> hardware computability list,' reasonable' comes to mind when looking at the
> supported cards.

Well, it looks like maybe they started by bringing the antique 2nd Edition
(1995) drivers forward, to get 3rd Edition off the ground.  And 3rd Edition
has only been out a few months.  Maybe this situation will improve.

These days, though, every graphics card for the PC is a 3D gaming card, and
right now Matrox is the only company doing the right thing with documentation.
You can look at XFree86, but XFree86 has a lot of problems with a lot of
cards; bugs are rife, and many drivers aren't even accelerated.  So it's a
pretty grim situation.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[Scott Schwartz]

At the cost of great ugliness, maybe it would be feasible to build an
interface that could use NT drivers.   Didn't someone do something
like that for mach?

Re: [9fans] Are nvidia-cards working with plan9?

[Rick Hohensee]

>
> At the cost of great ugliness, maybe it would be feasible to build an
> interface that could use NT drivers.   Didn't someone do something
> like that for mach?
>
>

If so, that would be "to Mach"

Rick Hohensee

Re: [9fans] Are nvidia-cards working with plan9?

[root]

In article <39D232D9.D3B50609@arl.army.mil>,
	"Douglas A. Gwyn" <gwyn@arl.army.mil> writes:

> By all means develop drivers for other cards that interest you,
> and contribute them to the cause.  If they work at all I suspect
> they'll be added to the distribution.

For this plan to work, people must have the system installed
and functional.  If they need a graphics card driver first,
that might be difficult.

Kind of like a dog chasing its own tail.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[Douglas A. Gwyn]

root@mail.c-zone.net wrote:
> For this plan to work, people must have the system installed
> and functional.  If they need a graphics card driver first,
> that might be difficult.

It's easy enough to install a supported graphics card.

Re: [9fans] Are nvidia-cards working with plan9?

[jiho]

In article <39D90BF6.9133377@null.net>,
	"Douglas A. Gwyn" <DAGwyn@null.net> writes:

> root@mail.c-zone.net wrote:
>> For this plan to work, people must have the system installed
>> and functional.  If they need a graphics card driver first,
>> that might be difficult.
>
> It's easy enough to install a supported graphics card.

Assuming you have (or can get) one.  Most of them are obsolete.

Then there's the issue of either setting up a spare machine,
or replacing the (better) card in the machine you use....


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[sah]

> > It's easy enough to install a supported graphics card.
>
> Assuming you have (or can get) one.  Most of them are obsolete.

We're using the Diamond Stealth 3d 2000 vers 1.04 here.  While
obsolete, there are workarounds.  You can actually get a DS3D2K with other
bios versions (higher than 1.04), pull the chip off, and stick in a 1.04
you've burned yourself and it works great (e-me for board revs we've
tested this with).  I think I found one of these cards on ebay a few
months back for around $12.

I'll pass along the MOT S19 records for the 1.04 bios if anyone would
like them.

Sam

Re: [9fans] Are nvidia-cards working with plan9?

[sah]

s/I'll pass along the MOT S19 records for the 1.04 bios if anyone would
like them//

Hrm.  It has been brought to my attention that this action could land me
some quality time with a lawyer.  As I rarely have time for this these
days, I'm printing a retraction.

Cheers,

Sam

Re: [9fans] Are nvidia-cards working with plan9?

[jiho]

In article <8qnit2$37p$1@supernews.com>,
	Conor <conoruser@lineone.net> writes:

> Is there something wrong with this newsgroup for it seems to take 12/24
> hours to post to it?

Perhaps someone screens every post.

[Moderator's note: the above is correct; comp.os.plan9 is a
 moderated Newsgroup.  See:

 http://www.ibiblio.org/usenet-i/info/format.html#M

 for brief details.  Denis McKeon's article, "Moderated Newsgroups
 FAQ" posted to the news.answers Newsgroup is useful reading.  Russ
 Allbery's article, "Pitfalls of Newsgroup Moderation" which is
 posted to the same Newsgroup is further useful reading.]


Whether that would be "something wrong", I won't venture.


--Jim Howard  <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[root]

In article <20000925110001.18FDE19A26@mail>,
	forsyth@caldo.demon.co.uk writes:

>>> I can't help but wonder where a lot of other OSes would have been
>>> right now if they had not turned their noses up at 'gaming hardware'.
>>> It's a good thing that

> it isn't a case of turning up noses at games or gaming hardware

3D visualization is one of the areas where distributed computing
still fills a real need.  So a 3D card in a PC isn't necessarily
just for games, with a system like Plan 9.


--Jim Howard <jiho@mail.c-zone.net>

Re: [9fans] Are nvidia-cards working with plan9?

[nigel]

[-- Attachment #1: Type: text/plain, Size: 979 bytes --]

In general, the more 3D oriented, and state-of-the-art,
the less likely the card is to be supported. This is partially
a function of the proprietary nature of 'gamers' cards,
and partially the unlikelihood that researchers in Plan 9
have found the need to support such cards.

If you read the hardware compatibility specs, you will
see what is supported in the current release. The usual
line is 'ATi mach64 based cards are OK'. Of course, if
you see it mentioned in /lib/vgadb then this is encouraging.

Support is improving, though. If you check the archives,
you will see that a Voodoo 3 and S3 Savage drivers have been
done, but not yet released. Matrox G200/G400 may not be far away.

On the flip side, no Ati 128 bit cards are supported.

Another important rule of thumb is 'does XFree86 support
it'. This is an indication as to whether the technical
documentation needed to write a driver is available.
At one time certainly, Nvidia were difficult.


[-- Attachment #2: Type: message/rfc822, Size: 1945 bytes --]

From: Henri Philipps <Henri.Philipps@web.de>
To: 9fans@cse.psu.edu
Subject: [9fans] Are nvidia-cards working with plan9?
Date: Thu, 21 Sep 2000 08:27:49 GMT
Message-ID: <7l5eq8.fn.ln@127.0.0.1>

Hi,

are graphic-boards with nvidia chipset working with plan9?

I have a GeForce2 and couldnt find a way to make it work by changing the
vgadb file. Can anyone give me a hint in this direction?

Where is the best place to find an answer to this kind of questions?

Thanks,

Henri

Re: [9fans] Are nvidia-cards working with plan9?

[MoJoJoJo]

Looks like nVidia cards are supported by XFree with hardware
acceleration in versions 3.3.6 and 4.0.1 -- see:
http://www.xfree86.org/4.0.1/Status21.html#21

The latest plan9 hardware support document is at:
http://plan9.bell-labs.com/plan9dist/pchardware.html

Is the ultimate goal of this OS to be just a developer test platform?
I can't help but wonder where a lot of other OSes would have been
right now if they had not turned their noses up at 'gaming hardware'.
It's a good thing that 

--TNT2 Ultra Owner

On Thu, 21 Sep 2000 09:14:14 GMT, nigel@9fs.org wrote:

>In general, the more 3D oriented, and state-of-the-art,
>the less likely the card is to be supported. This is partially
>a function of the proprietary nature of 'gamers' cards,
>and partially the unlikelihood that researchers in Plan 9
>have found the need to support such cards.
>
>If you read the hardware compatibility specs, you will
>see what is supported in the current release. The usual
>line is 'ATi mach64 based cards are OK'. Of course, if
>you see it mentioned in /lib/vgadb then this is encouraging.
>
>Support is improving, though. If you check the archives,
>you will see that a Voodoo 3 and S3 Savage drivers have been
>done, but not yet released. Matrox G200/G400 may not be far away.
>
>On the flip side, no Ati 128 bit cards are supported.
>
>Another important rule of thumb is 'does XFree86 support
>it'. This is an indication as to whether the technical
>documentation needed to write a driver is available.
>At one time certainly, Nvidia were difficult.

[9fans] Are nvidia-cards working with plan9?

[Henri Philipps]

Hi,

are graphic-boards with nvidia chipset working with plan9?

I have a GeForce2 and couldnt find a way to make it work by changing the
vgadb file. Can anyone give me a hint in this direction?

Where is the best place to find an answer to this kind of questions?

Thanks,

Henri