TI34020 discontinued

34 views
Skip to first unread message

Jim Bryan

unread,
Sep 9, 1993, 12:39:00 AM9/9/93
to
I learned today that Texas Instrument will soon discontinue the
manufacture if its 34020 graphics engine. As a result several
versatile graphics cards will cease to be available after the
first quarter of 1994. Does anyone know what device or devices
are likely to replace this relatively inexpensive graphics
workhorse.
Jim Bryan - jim....@satalink.com

Brian Hook

unread,
Sep 9, 1993, 6:05:10 AM9/9/93
to

As far as I can tell the reason for the discontinuance of the 340x0 line is
simply that they were not cost competetive with S3, P9000, etc. Most
people were buying 340x0 boards for either Windows or AutoCad, and in these
two environments the limited but highly specialized functions of the
Windows accelerators outstripped the speed of the costlier 340x0 based
boards. For example, the Diamond Viper (Weitek P9000) handily DESTROYS the
34020 in most Windows benchmarks I've seen -- and it costs around 450,
whereas 340x0 based boards start at 995 and often go into the 2-4K range.

Where the 340x0 based boards truly shined were in image processing and 3d
graphics, where their true capabilities (and true advantages) over the
Windows accelerators shone. Specifically, the ability to store data
locally on onboard DRAM, large amounts of VRAM, the ability to write and
upload code to the TI chip, availability of dedicated floating point unit
that could be stacked and run in parallel ( 34082 ), etc. but these
abilities just don't satisfy production requirements (I assume).

From what I've been hearing dedicated proprietary ASIC designs are coming
into play as replacements. Specifically, the SPEA boards, various i860
designs, Pelluid/SGI IrisVision, Artist Graphics' XJS-1000 GPX, Matrox,
etc. Unfortunately, each of these boards tends to be optimized for only
one area of graphics. The exception is the Matrox, which appears highly
versatile, with great Windows, Autocad, and real-time 3d capabilities.
Unfortuantely, since it is a custom chip that is also proprietary, 3rd
party support may be lacking. And it costs as much as a 34020 board.

Brian

--
//----------------------------------------------------------------------------
// Brian W.K. Hook "Stop! Stop in the name of all that
// ( b...@beach.cis.ufl.edu ) which does not suck!" - Butthead
// ( 72144,3662 CI$ )
//
// "Oooooooo-la-tec! Ooooooo-la-tec!"
//
// For info on my 3d graphics project:
// finger b...@beach.cis.ufl.edu | more
//----------------------------------------------------------------------------

David Farley

unread,
Sep 9, 1993, 4:00:33 PM9/9/93
to
I thought I'd use this thread as an opportunity to ask a couple stupid
questions. Awhile back I bought an Opta Monalisa card which uses the
34020 chip because I needed to do 24 bit graphics at a resolution of
1024x768 in ms-windows, and at the time it was the only affordable
solution. The card still works and I'm still using it, but Opta has
gone out of business. Does anybody know if a) anybody picked up
support for the Monalisa card or b) if some other cards more recent
TIGA driver will work with the Monalisa? The impression I've gotten
is that the generic TIGA drivers, like the one that comes with
ms-windows, work at resolutions of up to 1024x768 by 256 colors, but
anything after that is specific to the card and the custom drivers.
The drivers I have leave couple things to be desired, although I'm it
works well enough for what I do that I'm not about to run out and buy
an entirely new video card at the moment. Anybody else out there
still using an Opta card?
--

David Farley dg...@midway.uchicago.edu

Neil Gould

unread,
Sep 10, 1993, 11:28:55 AM9/10/93
to

In a previous article, dg...@ellis.uchicago.edu (David Farley) says:

>The drivers I have leave couple things to be desired, although I'm it
>works well enough for what I do that I'm not about to run out and buy
>an entirely new video card at the moment. Anybody else out there
>still using an Opta card?
>--

Yep, I've got one of those puppies, too. Many of your concerns are also
mine, though it isn't the first time I've been "orphaned" by using a top-
of-the-line product, and I think the benefits outweighed the difficulties.

When the thing crashes, or software becomes incompatible with it, I'll just
replace it and keep on truckin'. The way that I work, I don't tend to up-
grade systems beyond a certain point, anyway. It's easier and jsut as econ-
omical to build a new system from the ground up.

As for the Windoze drivers for the Opta, mine work well enough, and the
system is quite stable, so unless there is a hardware failure there isn't
a lot of motivation for change.

--
neil gould + aa...@cleveland.freenet.edu
electronic media producer + Terra Tu A\V
SysOp: Video Arts Department; Co-SIGop - Audio Arts SIG

Ian Sherlock

unread,
Sep 10, 1993, 5:33:57 PM9/10/93
to
jim....@satalink.com (Jim Bryan) writes:

I'd just like to point out that Texas Instruments has ** NO ** plans to
discontinue the manufacture of the 34020.

The 34020 continues to be used in X terminal and embedded applications.
You can call your local TI sales office for more information on the
34020.

Regards
Ian Sherlock
Texas Instruments
--


------------------------------------------------------------------------
Ian J. Sherlock Phone: (713) 274-2957

Tom Williamson

unread,
Sep 10, 1993, 5:58:27 PM9/10/93
to
jim....@satalink.com (Jim Bryan) writes:
: I learned today that Texas Instrument will soon discontinue the

Contrary to the information Jim has received, TI has not and does not plan
to discontinue the TMS340xx family in the forseeable future.

It is true that TI has admitted defeat in the Windows graphics accelerator
market, giving way to accelerators based on S3 and P9000 chips. However,
the TMS340xx family of processors is used in many products other than
Windows accelerator cards.

It is somewhat unfortunate that the most visible applications for the
TMS340xx has been in the Windows accelerator market. The trade press has
inaccurately assumed that the withdrawl of the TMS340xx product from the
Windows market also meant that TI was discontinuing the product family
alltogether. This is simply not true!

The company I work for, Northwest Digital Systems, offers a line of
hardware and software based on the TMS340xx family for X Window System
applications. We are very close to TI and have discussed this matter
at length with them. They are still VERY interested in TMS340xx based
X Window System applications.

As an example, NDS has a design based upon the TMS34020-40 that offers a
complete, high performance X terminal on a card. It measures 7"x7" and
draws less than 10 watts of power. Applications for this card (we call it
Wildcat) range from embedded graphics engines in OEM hardware to standalone
X terminal designs offering very good performance at a very low cost.
XStone performance for Wildcat is 65,000 running an 8 bit color server.
Wildcat is not a PC based card, it is a standalone design.

I cannot offer much help to those of you needing assistance with Windows
graphics accelerators. However, when it comes to running the X Window
System on the TMS340xx, NDS is in a position to help. We continue to market
this product family and are clearly looking for new opportunities. I
apologize if this is long winded and a little commercial, but I felt the
need to clear the air regarding the uncertain status of the TMS340xx family
as reported in the trade journals. Simply trying to separate rumor from
reality.

Tom Williamson
Director, Sales & Marketing
Northwest Digital Systems, Inc.

Tom Williamson | Northwest Digital Systems
UUCP: uunet!nds!tom | Seattle, Washington USA
Internet: t...@nds.com | Phone: 206-524-0014 FAX: 206-524-3440

Michael A. Wilson

unread,
Sep 10, 1993, 9:07:03 PM9/10/93
to

I realize that this probably comes up a lot over here, but is there a
useful functional form for generating RGB values as a function of the
wavelength of light.

Thanks for any information,
Mike
--
_____________________________________________________________________________
Mike Wilson, MS 239-4 mwi...@max.arc.nasa.gov (415) 604-5759 day
NASA Ames Research Center (415) 604-1088 fax
Moffett Field, CA 94035 (510) 523-7790 eve

Joseph J. Kubicky

unread,
Sep 11, 1993, 7:54:02 PM9/11/93
to
i...@tms390.micro.ti.com (Ian Sherlock) writes:

>jim....@satalink.com (Jim Bryan) writes:

>I'd just like to point out that Texas Instruments has ** NO ** plans to
>discontinue the manufacture of the 34020.

>Regards


>Ian Sherlock
>Texas Instruments
>--

...however, I belive TI IS discontiuing SUPPORT for these devices in the
form of development/emulation systems. Therefore, don't expect to see many
new products being released using the 340 GSPs. As far as I know, TI will
continue to sell/support ONLY THE SILICON - nothing else.

Jay Kubicky

PS: This is probably just as well. I recently had cause to debug a system
using the TI boundary-scan debug/emulation system, and it's pretty crappy -
it crashes a lot (read: crashes the HOST PC it's running on) and doesn't
even seem to let you directly assemble code into memory (even DEBUG lets
you do this).

Ian Sherlock

unread,
Sep 13, 1993, 6:05:02 PM9/13/93
to
jkub...@cco.caltech.edu (Joseph J. Kubicky) writes:

>i...@tms390.micro.ti.com (Ian Sherlock) writes:

>>jim....@satalink.com (Jim Bryan) writes:

>...however, I belive TI IS discontiuing SUPPORT for these devices in the
>form of development/emulation systems. Therefore, don't expect to see many
>new products being released using the 340 GSPs. As far as I know, TI will
>continue to sell/support ONLY THE SILICON - nothing else.

>Jay Kubicky

Here is a little more detailed information on TI's graphics strategy .

TI ON GRAPHICS : CONTINUED PRODUCTION AND NEW DEVELOPMENTS

In 1992, the TI 340 Coprocessor/TIGA family lost it's price/performance
advantage in the mainstream PC segment to the more targeted accelerated
SuperVGA products like S3 and ATI. As a result, 340 TIGA was focused into the
HIGH RESOLUTION, TRUE COLOR WINDOWS ACCELERATION and COLOR X-TERMINALS SEGMENTS
where it CLEARLY OFFERS LEADERSHIP PRICE/PERFORMANCE.

TI currently PLANS TO CONTINUE THE MANUFACTURE AND SALE OF MOST 34010 and
34020 (340X) SILICON AS LONG AS MARKET DEMAND WARRANTS. In addition, we are
offering a cost reduced version of 34020 to extend the life of 340 designs with
the intent of lowering the cost of the solution. Since it was clear that 340
extensions would not obtain leadership price/performance in the mainstream PC
market, we altered our graphics strategy . Although we will continue to
manufacture current products for the foreseeable future, we will not develop
and new 340 graphics coprocessors designs.

However, we did not get out of the graphics business, we merely focused
our efforts to emerging end equipment targeted solutions where TI could offer
competitive advantage.

First, TI continued to develop leadership high speed color palettes for
the high performance windows graphics accelerators. TI palettes are the
fastest in the industry and recommended by many of the Windows accelerator
controller manufacturers to accompany thier controllers.

Second, we launched our Multimedia DSP objective to address the growing
multimedia market. TI is the market leader in DSP and we are working to bring
our price/performance and DSP systems experience to the PC multimedia arena.

Furthermore, we continued development of leadership advanced processing
for the next generation of graphics and multimedia with our MVP (Multimedia
Video Processor). It is an advanced DSP that will debut in 1994 with initial
applications in high value programmable video systems such as video
conferencing systems. It should blow the doors off of any of it's competition.

Finally, we are pursuing an accelerated SVGA solution to tie in with our
PC end equipment requirements driven "Total Integration" thrust you will be
hearing more about in 1993.

Consequently, as you can see, TI is not abandoning the graphics market;
we are altering our product mix to better match target market requirements, as
they evolve, with competitive and leadership products.

Regards,
Ian Sherlock
Texas Instruments

Chris_F...@cup.portal.com

unread,
Sep 14, 1993, 8:31:19 AM9/14/93
to
Jeez... TI acts like the "PC market" is the ONLY market. What about those of
us who use TMS34020-based graphics accelerators on OTHER than PC platforms?
I don't suppose we get diddlysquat consideration when TI decides to pull the
plug (so to speak).

Chris Chiesa
Chris_F...@cup.portal.com

Neil Gould

unread,
Sep 14, 1993, 9:22:47 AM9/14/93
to

As a respondent to earlier parts of this thread, I'd like to point out
that the TIGA series of boards, including my "orphaned" Mona Lisa, are
_not_ the same types of products as the "Windoze Accelerators." For high-
resolution TrueColor applications, there are only a couple of products
outside of the TI series that deliver, and they are as pricey as the
TIGA boards, and as proprietary. One product offered a triple S3 solution,
and I haven't heard from them lately, either.

IMHO, hardware progress is going to make obsolete any graphics solution
well before there is decent software support, anyway. Since there are
very few uses for high resolution TrueColor in the mainstream market,
it isn't surprising that no new 340x series processors are being consi-
dered. New driver software for the existing processors would be real
nice! 8-)

Nick C. Fotis

unread,
Sep 14, 1993, 9:06:30 AM9/14/93
to
i...@tms390.micro.ti.com (Ian Sherlock) writes:

>jkub...@cco.caltech.edu (Joseph J. Kubicky) writes:
>>...however, I belive TI IS discontiuing SUPPORT for these devices in the
>>form of development/emulation systems. Therefore, don't expect to see many
>>new products being released using the 340 GSPs. As far as I know, TI will
>>continue to sell/support ONLY THE SILICON - nothing else.
>>Jay Kubicky

>Here is a little more detailed information on TI's graphics strategy .

>TI ON GRAPHICS : CONTINUED PRODUCTION AND NEW DEVELOPMENTS
>
> In 1992, the TI 340 Coprocessor/TIGA family lost it's price/performance
>advantage in the mainstream PC segment to the more targeted accelerated
>SuperVGA products like S3 and ATI. As a result, 340 TIGA was focused into the
>HIGH RESOLUTION, TRUE COLOR WINDOWS ACCELERATION and COLOR X-TERMINALS SEGMENTS
>where it CLEARLY OFFERS LEADERSHIP PRICE/PERFORMANCE.

A small question: what if I want (say) to make a 3D engine for a personal
computer like the Amiga or the Macintosh? Are there any cheap and fast
solutions that provide things like 3D transformations, clipping,
gouraud shading, etc.?? (Something like a low cost Geometry Engine)

Clearly, TI-based cards failed here (more than 2K USD/card is not affordable!).

Inquiring minds want to know.

Greetings,
Nick.
--
Nick (Nikolaos) Fotis National Technical Univ. of Athens, Greece
HOME: 16 Esperidon St., InterNet : nfo...@theseas.ntua.gr
Halandri, GR - 152 32 UUCP: mcsun!pythia!theseas!nfotis
Athens, GREECE FAX: (+30 1) 77 84 578

Brian Hook

unread,
Sep 14, 1993, 11:49:48 AM9/14/93
to
In article <nfotis.748011990@theseas> nfo...@ntua.gr (Nick C. Fotis) writes:

> A small question: what if I want (say) to make a 3D engine for a personal
> computer like the Amiga or the Macintosh? Are there any cheap and fast
> solutions that provide things like 3D transformations, clipping,
> gouraud shading, etc.?? (Something like a low cost Geometry Engine)

I don't know about Mac, but the Amiga had a couple 34020 boards that used
the incredible 34082 floating point unit. The 34020/34082 pair were really
nice, and TI had a graphics library available (with a great tutorial, buy
the way, on 3d graphics) for it that was FREE (the 34020 library wasn't
free, but the 082 library was). It did everything, including Gouraud
shading, clipping, etc. The 34082 had clipping/shading INSTRUCTIONS! And
you could code complex instructions and upload them to the 34082. And have
up to 7 (?) 34082s running parallel on a single 34020 board.

Basically, you could write an entire program in C, some parts in 34082
assembly (or use the library), upload it to the board, and your PC became a
real fast input device into the TI stuff. The 34082 wasn't that expensive
(I believe it was in the couple of hundred dollar range, at the time the
same cost as an Intel 387). And for hard core types, there were a couple
of board available with a 34020 for 2D rendering and an i860 for 3d
processing (YES!).

Kendall Bennett

unread,
Sep 16, 1993, 9:00:13 PM9/16/93
to
nfo...@ntua.gr (Nick C. Fotis) writes:

>A small question: what if I want (say) to make a 3D engine for a personal
>computer like the Amiga or the Macintosh? Are there any cheap and fast
>solutions that provide things like 3D transformations, clipping,
>gouraud shading, etc.?? (Something like a low cost Geometry Engine)

Keep your eyes open for the new 3D graphics accelerator solutions that
are appearing for Unix workstations and PC graphics systems. Du Pont
Pixel has a new GLiNT chip for doing OpenGL rendering that will be
incorporated into a number of cards for Unix workstations and PC's. There
are also a number of chips manufacturers in Taiwan working on 3D
graphics engines specially for this purpose, and will probably be
shipping samples some time next year.

Given that 3D is now making it into the mainstream, and the fact that
Windows 4.0 (Windows NT 2.0) _may_ have OpenGL support and OS/2 may
have OpenGL support (it already exists from special hardware vendors),
you will see the popularity of these things increase and the price
of the hardware drop in the coming years. Soon we just may well have
affordable 3D solutions on PC based systens...

+------------------------------------------+-------------------------------+
| Kendall Bennett | Internet: |
| RMIT Advanced Computer Graphics Centre | k...@citri.edu.au |
| CITRI Building, 723 Swanston Street | rc...@minyos.xx.rmit.oz.au |
| Carlton Victoria 3053 AUSTRALIA. | |
+------------------------------------------+-------------------------------+

Nick C. Fotis

unread,
Sep 17, 1993, 5:11:12 PM9/17/93
to

[ Note: I'm cross-posting to comp.sys.amiga.graphics also, as this takes a
more Amiga-oriented point of view -- nfotis ]

b...@beach.cis.ufl.edu (Brian Hook) writes:

>In article <nfotis.748011990@theseas> nfo...@ntua.gr (Nick C. Fotis) writes:

>> A small question: what if I want (say) to make a 3D engine for a personal
>> computer like the Amiga or the Macintosh? Are there any cheap and fast
>> solutions that provide things like 3D transformations, clipping,
>> gouraud shading, etc.?? (Something like a low cost Geometry Engine)

>I don't know about Mac, but the Amiga had a couple 34020 boards that used
>the incredible 34082 floating point unit. The 34020/34082 pair were really

>nice, and TI had a graphics library available (with a great tutorial, by


>the way, on 3d graphics) for it that was FREE (the 34020 library wasn't
>free, but the 082 library was). It did everything, including Gouraud
>shading, clipping, etc. The 34082 had clipping/shading INSTRUCTIONS! And
>you could code complex instructions and upload them to the 34082. And have
>up to 7 (?) 34082s running parallel on a single 34020 board.

^
Up to 4 was the limit on cards like the Vivid-24. There was the (well known)
claim from the company that the 4 FPUs could achieve 160 MFLOPS (Ha!), but
they "forgot" to add that it was the purely theoretical limit...

I don't know if the TI-based products could keep up with systems like an
i860 + custom VLSI for 3D work (if the guy from TI can help here... )

>Basically, you could write an entire program in C, some parts in 34082
>assembly (or use the library), upload it to the board, and your PC became a
>real fast input device into the TI stuff. The 34082 wasn't that expensive
>(I believe it was in the couple of hundred dollar range, at the time the
>same cost as an Intel 387). And for hard core types, there were a couple
>of board available with a 34020 for 2D rendering and an i860 for 3d
>processing (YES!).

All fine and dandy (as far as computing power goes), but you forgot that I
said *cheap and fast*. And to pay more and more wasn't a very pleasing
proposition. If (at least on the Amiga) there was a standard for TI-based
cards (I heard about a try with the so-called SAGE standard, but the
companies who made these cards nearly forgot the software support part :-( )


Not a serious Amiga user could develop for such a high-priced board (we say
prices like 3k USD + up!!) with a so low market mass (how many were sold?
One, two hundred??).
If there was a low cost entry price (less than 1,000 USD for a 24-bit capable
card in SGVA resolutions) and better software support, who knows?

Well, I suppose that the cost in the i860+34020 cards was, erm *very* high.....
(an i860 is more than adequate for me, as a part of an SGI's Geometry
Engine). Or someone could say that these cards were before their time
(paraphrased, we now have software which could put these cards to good use,
and these cards are now disappearing!!)

Someone told me that DSPs are capable of taking the role of a Geometry
Engine, but I doubt it very highly. These chips seem very primitive to
me for performing all these tasks the traditional graphics pipeline wants.

What about "porting" an SGI Geometry Engine to the 32-bit Amigas (with OpenGL,
etc.)???

(hint, hint)

Greetings,
Nick.
[ Followups directed to comp.sys.amiga.graphics ]

--
Nick (Nikolaos) Fotis National Technical Univ. of Athens, Greece
HOME: 16 Esperidon St., InterNet : nfo...@theseas.ntua.gr
Halandri, GR - 152 32 UUCP: mcsun!pythia!theseas!nfotis
Athens, GREECE FAX: (+30 1) 77 84 578

USENET Editor of comp.graphics Resource Listing and soc.culture.greece FAQ
NTUA/UA ACM Student Chapter Chair - we're organizing a small conference
in Comp. Graphics, call if you're interested to participate.

Jon Leech

unread,
Sep 17, 1993, 6:36:34 PM9/17/93
to
In article <nfotis.748300272@theseas>, nfo...@ntua.gr (Nick C. Fotis) writes:
|> Someone told me that DSPs are capable of taking the role of a Geometry
|> Engine, but I doubt it very highly. These chips seem very primitive to
|> me for performing all these tasks the traditional graphics pipeline wants.

It helps to know what people have actually done before speculating like
this. DSPs can do 3D graphics with no trouble, as witness (for example) the
AT&T Pixel Machine, based on the DSP32 chip. And DSP32s were a hell of a lot
easier to hand-code efficient assembler for than i860s. Considerations for
picking one or the other type of processor have to do with many other
factors than the "primitiveness" of the processor.

Jon
__@/

Reply all
Reply to author
Forward
0 new messages