Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

YARC and Apple 3D card comments?

1 view
Skip to first unread message

M.W.Derryberry

unread,
Feb 20, 1996, 3:00:00 AM2/20/96
to
> I am considering buying a 3D card for my new 9500. I understand that
> there are YARC and Apple 3D boards. Has anyone has experiences with these
> cards? Speed, ease of use, compatibility, etc...

Everything I have read about the Quickdraw 3D cards from Apple sounds
great. Supposedly it is the cheapest and provides the highest performance.

--mike


Kevin Stone

unread,
Feb 21, 1996, 3:00:00 AM2/21/96
to
M.W.Derryberry (mw...@esygvl.com) wrote:
: > I am considering buying a 3D card for my new 9500. I understand that


CRAP, CRAP, and more CRAP.. that's what best discribes the three QD3D
cards recently reviewed by MacWorld. The AppleQD3D card ($400), the Maxtrix
Millennium ($650), and the Yarc Screamer ($950).

MacWorld labs used 3D World 1.0 for thier tests. 3D World is a
$99, low end 3D modeling and visualization program that was designed
completely with the QD3D library. Apparently other programs that have
been retrofited with QD3D do not perform as well since they have many
features that are not accerated by the boards. So theoreticaly, 3D World
should show the maximum benefit of the QD3D accelerators.

The tests used three catagories to judge the speeds: One used a
5740 polygon scene, two used a smaller 3393 polygon scene, and the
third used the same 3393 polygon scene but with 4 small texture maps.
All scenes were Gouraud shaded.
The test computer was a standard 8500/120 with 32Megs of RAM.

-------------- tests results ------------
Apple's card came out on top showing a meger 20% improvement on
both non-texturemaped scenes. On the texturemaped scene, that shot up to
a 90% improvement over the processor alone. Pittiful when compared to
promises of 1200% improvements when the cards were annouced late last
year. Also pittiful next to standard QuickDraw graphics cards that
accelerate regular QuickDraw code up around 1500% above the processor.

Matrox Millennium cards showed only a 10% gain over the processor
on the non-texturemaped scenes. The card could not caculate the
texturemaped sences as it does not support texturemaping.

Yarc Systems, 'Yarc Screamer' yawned... showing dimsel
performance gains of 5% to 10% on the Gouraud shaded scenes. The
Screamer fell asleep durring the texturemaped tests, showing it to
calculate the scense up over 50% SLOWER than the processor alone.
-------------------------------------------

It appears from these tests that past promises of performance gains
of up to 12 times, are just not true... in fact, the Maxtrox and Yarc
cards can't even promise gains of even 12 percent in some instances. We
can only hope that the numbers displayed in these tests are either a fluke,
or that 3D world doesn't acctualy use the QD3D library after all. But,
all indications are that none of these cards.. even Apple's.. are worth
thier price for the less than mediocre performance they provide.

Sincerly,
-Kevin Stone

All informationed presented in this post was found in the 1996 April
issue of MacWorld Magazine, page 60.

John W. Blackburne

unread,
Feb 21, 1996, 3:00:00 AM2/21/96
to
In article <Pine.HPP.3.91.96022...@monaco.esygvl.com>,
"M.W.Derryberry" <mw...@esygvl.com> wrote:

:> I am considering buying a 3D card for my new 9500. I
understand that
:> there are YARC and Apple 3D boards. Has anyone has experiences with these
:> cards? Speed, ease of use, compatibility, etc...
:
:Everything I have read about the Quickdraw 3D cards from Apple sounds
:great. Supposedly it is the cheapest and provides the highest performance.

I'm not sure about performance: going by the quoted figures of triangles
per second most other boards, in particular GLING based boards, come out
faster than Apple's board. I've only see the Apple one in action though,
and there are no comparative reviews of the Mac cards to go by yet, so I'm
not sure how safe it is to go by these figures. Have a look at

http://www.matrox.com/mgaweb/macbroch.htm

and

http://www.yarc.com/glint.htm

for info and specs on the two shipping cards other than Apple's.

But price/performance wise the Apple card seems to come out best, as it's
about half the price of the other two boards. It's also more flexible, as
it will support acceleration on more than 1 display, and is scalable, so
you can get more acceleration by just adding a second card, and still pay
less than the Yarc card.

Then again the Matrox card is also a very nice video card, which is
getting good reviews in the PC press, and would be a good choice if you
wanted a flexible hi-res video card with accelerated 2D and 3D graphics. I
know little about the Yarc card apart from what's on their web page.

John
--
John Blackburne, jo...@tempest.net.hk
Programmer Asia, Inc. Online: http://www.asia-inc.com
Technology consultant and trainer: http://www.hk.super.net/~johnb

Larry Edwards

unread,
Feb 21, 1996, 3:00:00 AM2/21/96
to
Kevin Stone (kbs...@silver.sdsmt.edu) wrote:

: M.W.Derryberry (mw...@esygvl.com) wrote:
: : > I am considering buying a 3D card for my new 9500. I understand that
: : > there are YARC and Apple 3D boards. Has anyone has experiences with these
: : > cards? Speed, ease of use, compatibility, etc...
: :
: : Everything I have read about the Quickdraw 3D cards from Apple sounds
: : great. Supposedly it is the cheapest and provides the highest performance.

: CRAP, CRAP, and more CRAP.. that's what best discribes the three QD3D

: cards recently reviewed by MacWorld. The AppleQD3D card ($400), the Maxtrix
: Millennium ($650), and the Yarc Screamer ($950).

:
: [...]
: Sincerly,
: -Kevin Stone

: All informationed presented in this post was found in the 1996 April

^^^^^
: issue of MacWorld Magazine, page 60.

A little early for April fools isn't it?

Larry Edwards

Michael Shandony

unread,
Feb 22, 1996, 3:00:00 AM2/22/96
to
In article <4gg7a5$s...@nntp.stanford.edu>, Larry Edwards <edwards@finch> wrote:

>Kevin Stone (kbs...@silver.sdsmt.edu) wrote:
>: CRAP, CRAP, and more CRAP.. that's what best discribes the three QD3D
>: cards recently reviewed by MacWorld. The AppleQD3D card ($400), the Maxtrix
>: Millennium ($650), and the Yarc Screamer ($950).
>
>: All informationed presented in this post was found in the 1996 April
> ^^^^^
>: issue of MacWorld Magazine, page 60.
>
>A little early for April fools isn't it?

Only for those who bought one of these cards. It was in the April 1996
MacWorld. I just got mine two days ago. It gets kind of confusing
when publishers try to make their magazines sound more current by
putting future dates on their covers.

--
Michael Shandony | e-mail: michael....@eng.sun.com
2550 Garcia Avenue | phone: 415.786.6281
M/S: MPK16-202 | fax: 415.786.7957
Mountain View, CA 94043-1100 |

John W. Blackburne

unread,
Feb 22, 1996, 3:00:00 AM2/22/96
to
In article <4ge388$2...@news.sdsmt.edu>, kbs...@silver.sdsmt.edu (Kevin
Stone) wrote:

: Apple's card came out on top showing a meger 20% improvement on

:both non-texturemaped scenes. On the texturemaped scene, that shot up to
:a 90% improvement over the processor alone. Pittiful when compared to
:promises of 1200% improvements when the cards were annouced late last
:year. Also pittiful next to standard QuickDraw graphics cards that
:accelerate regular QuickDraw code up around 1500% above the processor.
:
: Matrox Millennium cards showed only a 10% gain over the processor
:on the non-texturemaped scenes. The card could not caculate the
:texturemaped sences as it does not support texturemaping.
:
: Yarc Systems, 'Yarc Screamer' yawned... showing dimsel
:performance gains of 5% to 10% on the Gouraud shaded scenes. The
:Screamer fell asleep durring the texturemaped tests, showing it to
:calculate the scense up over 50% SLOWER than the processor alone.

The problem is that most traditional apps do not seem to make good use of
QD3D acceleration. 3D world & Strata Studio Pro Blitz both use QD3D and
will benefit from an accelerator when it's there, but the gains are not
spectacular. Without sitting down with the application code it's difficult
to know why, but I would guess each app just spends too much time in non
QD3D routines, either in the app itself or by giving other apps a lot of
time, and such code would not benefit from acceleration. Apps using the
QD3D viewer, such as SimpleText and the Scrapbook, also do not seem to
gain dramatically from HW acceleration.

The apps that I've seen most benefit are apps which spend almost all their
time in QD3D routines but don't use the QD3D viewer, typically simple apps
with a limited human interface. E.g. Gerbils, TextureEyes (both available
from Apple's QD3D pages), NewEra (a demo app on the QD3D book CD), Go Cart
CD, TicTacToe3D and Wormhole3D (all by Brian Greenstonem and avaiilable
from info-mac etc.).


All of these are speeded up many fold by Apple's QD3D accelerator,
typically going from a jerky 3-10 fps to a smooth 30+ fps. Some also take
advantage of an accelerator's other capabilties, i.e. CSG and tranparency,
without slowing down the acceration of general geometry, something
probably imnpossible to achieve in software in realtime.

Based on my own experiences Apple's claim of up to 12 x acceleration seems
in the right ballpark. Based on my own pretty unscientific observations I
would happily say I'm seeing about a 10 x speedup in a number of apps, and
these are apps that hardly push the limit of what QD3D and QD3D
acceleration can do.

Larry Edwards

unread,
Feb 22, 1996, 3:00:00 AM2/22/96
to
John W. Blackburne (jo...@hk.super.net) wrote:
: In article <4ge388$2...@news.sdsmt.edu>, kbs...@silver.sdsmt.edu (Kevin
: Stone) wrote:

: : Apple's card came out on top showing a meger 20% improvement on
: :both non-texturemaped scenes. On the texturemaped scene, that shot up to
: :a 90% improvement over the processor alone. Pittiful when compared to

: [...]
: The problem is that most traditional apps do not seem to make good use of


: QD3D acceleration. 3D world & Strata Studio Pro Blitz both use QD3D and
: will benefit from an accelerator when it's there, but the gains are not
: spectacular. Without sitting down with the application code it's difficult

[...]
: All of these are speeded up many fold by Apple's QD3D accelerator,


: typically going from a jerky 3-10 fps to a smooth 30+ fps. Some also take
: advantage of an accelerator's other capabilties, i.e. CSG and tranparency,
: without slowing down the acceration of general geometry, something
: probably imnpossible to achieve in software in realtime.

[...]

What parts of the rendering process does Apple's QD3D accelerator
handle... the MacWorld article was a little vague. Does the card do:

a) projection/viewing/model matrix concatenations
b) projection of vertices from 3D to 2D
c) rasterization of the projected polygon

If the main CPU is still handling some of the above then I could see
how performance improvements might vary wildly depending on the task.

Larry Edwards


Larry Edwards

unread,
Feb 22, 1996, 3:00:00 AM2/22/96
to
Michael Shandony (mic...@hotlegs.eng.sun.com) wrote:
: [...]
: >A little early for April fools isn't it?

: Only for those who bought one of these cards. It was in the April 1996
: MacWorld. I just got mine two days ago. It gets kind of confusing
: when publishers try to make their magazines sound more current by
: putting future dates on their covers.

Yep, I just checked their online version and read the story... pretty
disappointing perf., but one wonders if it's the failure of the app
programmers or the cards. Then again, one wonders if it's that hard to
get good performance, maybe it's the card...


Larry Edwards


Eric Kornblum

unread,
Feb 22, 1996, 3:00:00 AM2/22/96
to
I checked out all three boards at MacWorld expo in January. I was most
impressed by the Apple QD3D board. It seemed to give the best performance
gains. I even saw it running Havoc, which is a good tank-like game that
supports QD3D and it looked awesome on the Apple board. But, as the
reviews have shown, the performance gains aren't near what was claimed.
However, a new version of QD3D (version 1.1, I believe) is supposed to be
released this spring and it should help performance. Plus, let's give
these developers a break. These are the first EVER QD3D boards. We can't
expect them to deliver the kinds of performance gains that 2D QD boards
that have been around for years show. It's going to take a while for both
the hardware & the software to become polished. What we have now looks
like a good beginning though, & I plan on buying one of Apple's boards (as
soon as I get my 8500 : ) ).

John W. Blackburne

unread,
Feb 23, 1996, 3:00:00 AM2/23/96
to
In article <4gijq5$h...@nntp.Stanford.EDU>, edwards@finch (Larry Edwards) wrote:

:What parts of the rendering process does Apple's QD3D accelerator


:handle... the MacWorld article was a little vague. Does the card do:
:
:a) projection/viewing/model matrix concatenations
:b) projection of vertices from 3D to 2D
:c) rasterization of the projected polygon

As best I can tell the card handles all rendering tasks. Without breaking
into the debuggger and disassembling the code it's difficult to know
precisely, but I would think (b) and (c) are among the task it
acccelerates.

The math is mostly handled by the processor as far as I can tell. This is
not a big problems as most of the work is in the rendering and the PPC is
not math slouch - in particular it's 64 bit fp architecture and it's
support for multiply-&-add intructions means it is well suited to QD3D
matrix and vector math (& vice versa).

:If the main CPU is still handling some of the above then I could see


:how performance improvements might vary wildly depending on the task.

I doubt it's the QD3D code, but other considerations such as how often a
program calls WNE or gets user events and how many other non QD3D tasks it
is handling at the same time. Also apps may use the extra power to improve
in areas other than raw performance - e.g. Havoc if I remember correctly
gets faster and also generates much more impressive rendering.

Pablo Fernicola

unread,
Feb 23, 1996, 3:00:00 AM2/23/96
to
In article <4gijq5$h...@nntp.Stanford.EDU>, edwards@finch (Larry Edwards)
wrote:
>
>:What parts of the rendering process does Apple's QD3D accelerator
>:handle... the MacWorld article was a little vague. Does the card do:
>:
>:a) projection/viewing/model matrix concatenations
>:b) projection of vertices from 3D to 2D
>:c) rasterization of the projected polygon

Most, if not all, 3D accelerator plug-in cards for personal computers in
the market today only deal with the rasterization layer (c: in the list
above). This includes all QuickDraw 3D accelerators in the market today.
There are some computers that come with 3D on the motherboard which
perform higher level functions, but these machines are not usually in the
consumer space (their functionality is on par with Unix workstations and
they are priced at the same level).

Having said that, the QuickDraw 3D architecture allows any board
manufacturer to accelerate the 3D rendering pipeline at higher levels.
It's just that nobody has released a product for the consumer space yet.

it is important to note that there are operations which can be performed
at the rasterization level that provide high end functionality. Apple's
3D accelerator provides transparency, tri-linear mipmapped texture
mapping, and Constructive Solid Geometry (CSG) operations, which
previously were only available on high end workstations (and certainly
not at the $399 level).

Pablo Fernicola- p...@apple.com -QuickDraw 3D Tech Lead
----- This posting represents my personal opinions.-----
----- I do not represent/speak for my employer. -----

Phillip Mills

unread,
Feb 27, 1996, 3:00:00 AM2/27/96
to
In article <1996Feb23.1...@gallant.apple.com>, Pablo Fernicola
<p...@apple.com> wrote:

> Having said that, the QuickDraw 3D architecture allows any board
> manufacturer to accelerate the 3D rendering pipeline at higher levels.
> It's just that nobody has released a product for the consumer space yet.

It will be interesting to see what other companies will bring out 3D
cards. I know Creative (the soundblaster people) had there's but who
knows what happened to it (and there demos were all on video and not the
machine).

I can see the actual 3D processors getting more intergrated into the
actual logic board of pc's in the near future as VRML and all the 3D stuff
people are getting into, is getting more and more public.

-- Disclaimer:
Comments are all mine and have NOTHING to do with the company I work for...

Minister for Cutting Edge Technologies that Suck

unread,
Feb 27, 1996, 3:00:00 AM2/27/96
to
In article <4gg7a5$s...@nntp.Stanford.EDU>, edwards@finch (Larry Edwards) writes:
> Kevin Stone (kbs...@silver.sdsmt.edu) wrote:
> : M.W.Derryberry (mw...@esygvl.com) wrote:
> : : > I am considering buying a 3D card for my new 9500. I understand that
> : : > there are YARC and Apple 3D boards. Has anyone has experiences with these
> : : > cards? Speed, ease of use, compatibility, etc...
> : :
> : CRAP, CRAP, and more CRAP.. that's what best discribes the three QD3D
> : cards recently reviewed by MacWorld. The AppleQD3D card ($400), the Maxtrix
> : Millennium ($650), and the Yarc Screamer ($950).
>
> : All informationed presented in this post was found in the 1996 April
> : issue of MacWorld Magazine, page 60.
>
> A little early for April fools isn't it?

Um no, the April edition's already on-line. From what I remember they thought
the apple card was ok but the Yark one was a waste of money. For what it's
worth make sure you have a PCI mac with a system enabler version 1.1 or higher.
Otherwise the card won't work, and it's cdrom doesn't have the latest versions
of the os on it. You have been warned.

dave
------------------------------------------------------------------------
D...@geo.ed.ac.uk Dave Fitch
Department of Geography
University of Edinburgh
Scotland


Michael Eilers

unread,
Feb 27, 1996, 3:00:00 AM2/27/96
to
In article <fleep-27029...@17.32.67.164>, fl...@mpx.com.au
(Phillip Mills) wrote:

> In article <1996Feb23.1...@gallant.apple.com>, Pablo Fernicola
> <p...@apple.com> wrote:
>
> > Having said that, the QuickDraw 3D architecture allows any board
> > manufacturer to accelerate the 3D rendering pipeline at higher levels.
> > It's just that nobody has released a product for the consumer space yet.
>
> It will be interesting to see what other companies will bring out 3D
> cards. I know Creative (the soundblaster people) had there's but who
> knows what happened to it (and there demos were all on video and not the
> machine).
>

Supposedly Newer technologies (of accelerator/cache card fame) has one in
the works. might be interesting...

> I can see the actual 3D processors getting more intergrated into the
> actual logic board of pc's in the near future as VRML and all the 3D stuff
> people are getting into, is getting more and more public.

I fully expect the next generation of Powermacs (620?) to have 3-d DSPs
welded on the motherboard like SGIs--With a technology like Quickdraw 3D,
apple would be insane not to give it direct hardware support, and
aggresively broaden its base.

michael

bayleyp

unread,
Mar 3, 1996, 3:00:00 AM3/3/96
to
In article <4h39fo$l...@nntp.Stanford.EDU>, edwards@finch (Larry Edwards) wrote:

> Oh one thing I meant to ask, has anybody developed an OpenGL api
> extension exist for QD3D?
>
> Larry Edwards

Why? It's not like AudoDesk is making anything for the mac anyhow.
OpenGL is another example of SGI trying to make a standard "weak" enough
for the PC, and a bad standard like that. Another is VRML which is bloated
with info nobody has any use for.

bayleyp

unread,
Mar 3, 1996, 3:00:00 AM3/3/96
to
In article <johnb-29029...@annex6-57.hk.super.net>,

jo...@hk.super.net (John W. Blackburne) wrote:

> In article <fleep-27029...@17.32.67.164>, fl...@mpx.com.au
> (Phillip Mills) wrote:
>
> :In article <1996Feb23.1...@gallant.apple.com>, Pablo Fernicola
> :<p...@apple.com> wrote:
> :
> :> Having said that, the QuickDraw 3D architecture allows any board
> :> manufacturer to accelerate the 3D rendering pipeline at higher levels.
> :> It's just that nobody has released a product for the consumer space yet.
> :
> :It will be interesting to see what other companies will bring out 3D
> :cards. I know Creative (the soundblaster people) had there's but who
> :knows what happened to it (and there demos were all on video and not the
> :machine).
>

> Creative's (3D Blaster) card is VLBUS rather than PCI, so I doubt you'll
> see a Mac version of it any time soon. all other card's I've come across
> are PCI cards, so there's no reason why there couldn't be Mac versions of
> them.
>

3D labs are working on a professional PCI based 3D accelerator. They
say they are working on QD3D RAVE drivers (although I can't remember the
WWW page I saw it on).

John W. Blackburne

unread,
Mar 4, 1996, 3:00:00 AM3/4/96
to
In article <bayleyp-0203...@slip173.ucs.orst.edu>,
bay...@slip.net (bayleyp) wrote:

: 3D labs are working on a professional PCI based 3D accelerator. They


:say they are working on QD3D RAVE drivers (although I can't remember the
:WWW page I saw it on).

Perhaps
http://www.3Dlabs.com/app_permedia.htm
- well worth checking out, as was the first site (except for perhaps
MacWeek) to mention QD3D Rave, and has some good info and links on Mac 3D
boards. But 3D Labs are only making the chips, and are leaving it up to
others to make the boards.

Larry Edwards

unread,
Mar 4, 1996, 3:00:00 AM3/4/96
to
bayleyp (bay...@slip.net) wrote:

: In article <4h39fo$l...@nntp.Stanford.EDU>, edwards@finch (Larry Edwards) wrote:

: > Oh one thing I meant to ask, has anybody developed an OpenGL api
: > extension exist for QD3D?
: >
: > Larry Edwards

: Why? It's not like AudoDesk is making anything for the mac anyhow.

Not sure what Autodesk has do with it... the reason I ask is that I'd
like to take code I wrote on an SGI and port it to the Mac with a
minimum of fuss.

: OpenGL is another example of SGI trying to make a standard "weak" enough


: for the PC, and a bad standard like that.

Hmm, actually OpenGL is just a cleaned up version of the Iris GL
without the window management hooks. Other than the window management
stuff (and some display list editing commands) there's nothing in Iris
GL that you can't do in OpenGL... and there are some things in OpenGL
that aren't supported on some platforms in Iris GL. I've worked with a
variety of 3D graphics apis such as CORE, Phigs, PEXLib, IBM's PGL,
and I'd take OpenGL over any of those anyday. QD3D is not bad
... maybe suffering a little from featuritis, but it's a reasonable
api.

: Another is VRML which is bloated with info nobody has any use for.

Such as...

Larry Edwards

0 new messages