SGI Indigo: Dig. Audio & Animation
Bill Laut
Many moons ago, I posted an article in "rec.music.synth", regarding an
upcoming product that turns an SGI Indigo into disk-based digital audio
recorded and editor.
There are two mechanisms I have found so far. If I am wrong in
either, please correct me.
The first is a product made by WaveFrame Corporation, and is called
"TidalWave." It functions as a complete four-track recorder/editor, and
includes scrubbing, syncing, etc. It might also have console automation.
I don't know if this is part of the -demo- package that SGI offers,
but I saw one utility on the Indigo called "MultiTrack," and you guessed it!,
it's a four-track audio recorder/mixer. I haven't had time to extensively
play with it (to determine things like console automation, SMTPE lockup,
etc.), but it can do the following:
1) Supports four tracks simultaneously
2) Handles Analog and Digital Inputs
3) Supports "sound files" as a compliment to the audio inputs
4) Has basic EQ and compression
<<<CAUTION: BLATANTLY-BIASED, HIGHLY ARROGANT EDITORIAL FOLLOWS>>>
One of my overpowering ambitions over the last thirteen years has been to
make feature-length animated motion pictures that metaphorically expresses
the Love of God therapeutically to the audience, in the form of healing and
nurturing. Realizing that I am a nobody with no proven track record at
making movies, and considering the expense involved in making animated films
the traditional way, I chose to patiently wait for computer technology to
"catch up" with my vision, and then pursue the animation via scene simulation
techniques.
After having had the opportunity to play with an Indigo at work, I can
confidently state that my waiting is over!
The SGI Indigo is the closest thing I have yet seen, that gives you the raw
materials needed to construct an animation facility in the privacy of one's
home, and with a small compliment of outboard peripherals, has the inherent
resources to actually allow you to create a feature-length animated film, as
a solo or near-solo effort.
Finally, with its analog and digital audio capabilities, the Indigo becomes
the first *truely* affordable disk-based digital audio recorder/editor, that
with a little bit of custom software, can be linked up with your animation
to do complete post-production work!
And best of all, it's at a price that nearly everyone can afford!!!
There may be other boxes out that that can do the same thing, but until I
see them, the SGI Indigo gets my vote as being "the greatest thing since
sliced bread."
<<End of Editorial>>
One of the criticisms I read in the media about the Indigo is that its audio
electronics, at present, is not adequate for doing CD-quality work.
On my stereo system, I have a Sony PCM-601ESD processor, and believe me, it's
easily CD quality. It's a consumer version of their professional PCM audio
processors that are used in final CD mastering. As such, it's basically a DAT
recorder, but without the tape transport: It digitizes the audio at 44.1kHz,
and then produces both a composite video signal and an AES/EBU digital stream.
It also has a simultaneous output circuit, that can take either the composite
video or digital stream, and convert it back into the analog audio.
If I can get the client site to let me bring it in, I plan to hook my Sony
up to the Indigo, in order to use it to feed their MultiTrack recorder, and
play back what MultiTrack mixes down. If anyone is interested, I'll post
the results of my experiments here.
BTW, I recently received my licensed distribution of the BRL-CAD software.
It's a raytracing package that's ->FREE<- over the Internet, or only $500
if you want a distribution kit. I have successfully loaded it onto an
Indigo with only a few minor glitches (dealing with STDIN and STDOUT), and
even without the documentation I was able to get nice raytraced
simulations. As for speed, what normally would take three hours on a
VAX-11/780 took less than 30 minutes on the Indigo. A typical image of
512x512 was traced in under five minutes, usually three.
I know what I'm asking Santa for, this year... (:-)
--
Bill Laut Internet: la...@alien.gici.com
Gull Island Consultants, Inc. Phone: (616) 780-3321
Muskegon, MI 49440 >> "Usual disclaimers, apply within" <<
"Your majesty, I have a message from God for you." -- Judges 3:20
Jon Wätte
the first *truely* affordable disk-based digital audio recorder/editor, that
with a little bit of custom software, can be linked up with your animation
to do complete post-production work!
And best of all, it's at a price that nearly everyone can afford!!!
There is nothing the Indigo can do (sound-wise) that a Mac Quadra
can´t do better with a plug-in card (i.e. Pro Tools) and the
software available for the mac is SOOO much cheaper.
Not to mention the support for video and time-syncing that´s built-into
QuickTime, something the equivalent of which no other platform has come
close to.
simulations. As for speed, what normally would take three hours on a
VAX-11/780 took less than 30 minutes on the Indigo. A typical image of
512x512 was traced in under five minutes, usually three.
Uh? Not faster? The 11-780 is pretty lame for integer math, and still
pretty slow with floating point. A 33 MHz 68040 (as found in a Q950
or a NeXT) is actually faster for floating point than the R3000 found
in the Indigo. The Indigo is faster for integer math, though (but not
by a whole lot!)
Now, consider a full-length feature; that´s 1,5 hours at 24fps, meaning
129600 frames. If each frame takes 3 minutes to generate, that´s
270 _days_ just for the frame generation. Add time to animate and
try things out, and you´ll soon run out of steam...
--
-- Jon W{tte, h...@nada.kth.se, Mac Hacker Suedoise (not french speaking) --
The word "politics" is derived from the word "poly", meaning
"many", and the word "ticks", meaning "blood sucking parasites".
-- Larry Hardiman
Chris Trimble
>There is nothing the Indigo can do (sound-wise) that a Mac Quadra
>can4t do better with a plug-in card (i.e. Pro Tools) and the
>software available for the mac is SOOO much cheaper.
The thing that makes the Indigo truely great is that it has the computing
power to additive synthesis on the fly for who knows how many tracks. At
SigGraph I talked to a guy from Blue Note SoundWorks, who makes Bars and
Pipes for the machine. Even though there are something like 12 voices to
the machine, there are really an unlimited number since the waveforms are
just computed on the fly.
>Not to mention the support for video and time-syncing that4s built-into
>QuickTime, something the equivalent of which no other platform has come
>close to.
SGI has licensed QuickTime. This doesn't really belong on this group, but
let me digress for .03 seconds. QuickTime doesn't have anything to do with
platform. QuickTime is software, not hardware. If you want to talk about
time-syncs in hardware, there are still a lot of platforms (SGI included)
which can support it, and do so quite well.
>Uh? Not faster? The 11-780 is pretty lame for integer math, and still
>pretty slow with floating point. A 33 MHz 68040 (as found in a Q950
>or a NeXT) is actually faster for floating point than the R3000 found
>in the Indigo. The Indigo is faster for integer math, though (but not
>by a whole lot!)
WHAT? Last I heard the 33 Mhz 040 did about 2 MFLOPS. The
R3000 does 4.2. Also, just think about it : a 33 mhz RISC vs. a 33mhz
CISC. I know of no RISC that would be slower than a CISC at FLOPS if
they're both the same speed like that.
- Chris
--
Chris Trimble | Let me hear you make decisions
chr...@uiuc.edu | without your television.
Beckman Mac Support | Let me hear you speaking just for me. - Martin Gore
Lee Clarke
>la...@alien.gici.com (Bill Laut) writes:
>
> the first *truely* affordable disk-based digital audio recorder/editor, that
> with a little bit of custom software, can be linked up with your animation
> to do complete post-production work!
>
> And best of all, it's at a price that nearly everyone can afford!!!
>
>There is nothing the Indigo can do (sound-wise) that a Mac Quadra
>software available for the mac is SOOO much cheaper.
>
>QuickTime, something the equivalent of which no other platform has come
>close to.
>
> simulations. As for speed, what normally would take three hours on a
> VAX-11/780 took less than 30 minutes on the Indigo. A typical image of
> 512x512 was traced in under five minutes, usually three.
>
>Uh? Not faster? The 11-780 is pretty lame for integer math, and still
>pretty slow with floating point. A 33 MHz 68040 (as found in a Q950
>or a NeXT) is actually faster for floating point than the R3000 found
>in the Indigo. The Indigo is faster for integer math, though (but not
>by a whole lot!)
>
>129600 frames. If each frame takes 3 minutes to generate, that4s
>270 _days_ just for the frame generation. Add time to animate and
>
>--
> -- Jon W{tte, h...@nada.kth.se, Mac Hacker Suedoise (not french speaking) --
> The word "politics" is derived from the word "poly", meaning
> "many", and the word "ticks", meaning "blood sucking parasites".
> -- Larry Hardiman
The SGI Indigo blows away a Mac Quadra. It is as simple as that.
You are comparing apples to oranges when you compare the Quadra's video
and imaging capabilities with the IRIS Indigo's video and imaging capabilites.
An Indigo does not use the cpu to process image data the
way the Quadra does.
The Indigo has its own graphics and imaging engine. The Quadra does
not. Period. Display controllers don't count here. The Quadra uses more
cpu overhead manipulating images than does the Indigo's cpu, so the
Indigo's cpu can afford to be slower. But that is moot with the MIPS
R4000.
The Indigo's graphics engine can render 3D graphics, movie thru memory-
resident images. Combine them with sound via the sound card that is
more powerful than the Quadra's, or with direct-to-disk audio, and
drop in a video card for vcr recording/playback, and you have a multimedia
platform that I believe is comparable to the Quadra's price, with extra
extra capabilities the Quadra does not have.
The Indigo does have an FPU processor, the R4000 is now the MIPS cpu
of choice for the machine, and the Indigo can be configured in the XS-24,
Elan, and fitted with the Reality Engine that graduates the Indigo
from the multi-media/science/engineering platform and into a Virtual
Reality platform.
I'm not knocking the Quadra. It is an excellent multimedia platform.
But the SGI is a more powerful multimedia and visualization and imaging
platform. It should be. It was designed to be a visualization platform,
not a general-purpose workstation.
We have have both and I can make a direct comparison of capabilites
here.
I posted capabilities of the SGI's sound processor card sometime this
spring or summer. If anyone has saved that posting, you may want to
repost it.
The SGI sound card has dual DA and dual AD for true stereo sound.
It can sample and play back up to 48kHz stereo sound.
It can also monitor one sound source while recording another.
You have to buy an extra card for the Quadra to get that same capabilty.
The control panel for the audio card and multitrack, a digital sound
recorder and editor, come with IRIX, as does movie, imgworks, etc.
True, there is no QuickTime on the SGI, but give the 3rd party vendors
time. All the hardware is there and then some.
Personally, I would own both machines. But I would find MANY more uses
for the Indigo.
Question: Is there software for the Indigo similar to MacroMind MediaMaker
with VLAN and MIDI capabilties?
Lee Clarke
lcl...@oasys.dt.navy.mil
"Opinions are my own"
Lee Clarke
SGI did acquire the license.
Lee Clarke
lcl...@oasys.dt.navy.mil
"Opinions are my own."
Archer Sully
*
* (lots of nice things deleted for brevity) (thanks much!)
*
*One of the criticisms I read in the media about the Indigo is that its audio
*electronics, at present, is not adequate for doing CD-quality work.
*
The main problem with the Indigo audio section is not the electronics them-
selves, but the environment that that the live in. The inside of a computer
is a hostile place for analog and hybrid parts that are extrememly noise
sensitive. In particular, the Indigo sounds awful (lots of humm and noise)
if the grounding isn't absolutely perfect. I have found that using plugging
everything in through the same surge protector works reasonably well at home.
You mileage can (and will) vary.
Even with perfect grounding, I freely admit to not having full qualifications
as to judging the CD-qualitiness of the audio. I will leave that to the
sound recording professionals.
*
*--
*Bill Laut Internet: la...@alien.gici.com
Archer Sully <arc...@sgi.com> | A vote in the 1992 presidential election
| is not unlike a self-inflicted wound.
Archer Sully
*There is nothing the Indigo can do (sound-wise) that a Mac Quadra
*can4t do better with a plug-in card (i.e. Pro Tools) and the
*software available for the mac is SOOO much cheaper.
*
Hmmm?? Most of the software is included on the machine, not to mention
an OS that allows for better integration of sound, not to mention sound
being standard so that all applications can use it...
*Not to mention the support for video and time-syncing that4s built-into
*QuickTime, something the equivalent of which no other platform has come
*close to.
*
Chuckle.
* simulations. As for speed, what normally would take three hours on a
* VAX-11/780 took less than 30 minutes on the Indigo. A typical image of
* 512x512 was traced in under five minutes, usually three.
*
*Uh? Not faster? The 11-780 is pretty lame for integer math, and still
*pretty slow with floating point. A 33 MHz 68040 (as found in a Q950
*or a NeXT) is actually faster for floating point than the R3000 found
*in the Indigo. The Indigo is faster for integer math, though (but not
*by a whole lot!)
Prove it. I can think of reasons why his ray trace was slow that are quite
independant of CPU speed, such as main memory size.
*
*Now, consider a full-length feature; that4s 1,5 hours at 24fps, meaning
*129600 frames. If each frame takes 3 minutes to generate, that4s
*270 _days_ just for the frame generation. Add time to animate and
*try things out, and you4ll soon run out of steam...
Now you start to have a point, but again, it depends on what sort of animation
we are talking about. If he's planning to draw every frame (or series of
frames) in 2D, then rendering time is unimportant. However, if he's working
in 3D, and animating and rendering frames, that's a totally different story.
SGI built its reputation on machines that are/were designed for fast modeling
of animation, and can do some fairly respectable rendering in real time.
This sort of capability is not supported on the Quadra.
An interesting side note. Earlier this year at MacWorld (?) there was a
Quadra vs. Indigo shootout. A Quadra was loaded up with all sorts of 3rd
party gear to get into the same league as an Indigo (albeit at 2x cost),
and sent head-to-head with an Indigo. Power up, and the Quadra smoked.
Turned out that there wasn't enough power from the power supply for everything
they had plugged in. An fx was brought in as the second string, and lost
handily in all speed tests.
Archer Sully
*The Indigo does have an FPU processor, the R4000 is now the MIPS cpu
*of choice for the machine, and the Indigo can be configured in the XS-24,
*Elan, and fitted with the Reality Engine that graduates the Indigo
*from the multi-media/science/engineering platform and into a Virtual
*Reality platform.
*
The Reality Engine has more processors than the Indigo has chips, (eg
it doesn't fit). Not only that, but it only fits in the larger chassis,
and costs as much as 3 Elans!
*
*True, there is no QuickTime on the SGI, but give the 3rd party vendors
*time. All the hardware is there and then some.
*
Its been licensed, and will ship someday ;-).
*Personally, I would own both machines. But I would find MANY more uses
*for the Indigo.
*
*Question: Is there software for the Indigo similar to MacroMind MediaMaker
*with VLAN and MIDI capabilties?
Not yet.
Mark Zuber
1. MFLOPS mean ABSOLUTELY NOTHING AT ALL!!!!
2. It takes more instructions in RISC to equal some of the instructions
in CISC, so performance calculations (like MFLOPS and MIPS)
get screwed up in the different instruction count values.
3. Most important, performance GREATLY depends on the PROGRAMMER of the
machine. Put a non-programmer on a RISC machine and a GREAT programmer
on a Commodore 64, and I'll bet the C-64 programmer could find a way
to make his C-64 program look faster.
\
Mark Zuber
Sr. in Electrical and Computer Engineering
Oregon State University
--
/-------------------------------------------------------------\
| Mark Zuber zub...@hobo.ece.orst.edu ATO - VTL |
| "A conclusion is the place where you got tired of thinking" |
\-------------------------------------------------------------/
Chris Wilson
>d88...@alv.nada.kth.se (Jon Wdtte) writes:
>>There is nothing the Indigo can do (sound-wise) that a Mac Quadra
>>can4t do better with a plug-in card (i.e. Pro Tools) and the
>>software available for the mac is SOOO much cheaper.
>>Uh? Not faster? The 11-780 is pretty lame for integer math, and still
>>pretty slow with floating point. A 33 MHz 68040 (as found in a Q950
>>or a NeXT) is actually faster for floating point than the R3000 found
>>in the Indigo. The Indigo is faster for integer math, though (but not
>>by a whole lot!)
>
> WHAT? Last I heard the 33 Mhz 040 did about 2 MFLOPS. The
> R3000 does 4.2. Also, just think about it : a 33 mhz RISC vs. a 33mhz
> CISC. I know of no RISC that would be slower than a CISC at FLOPS if
> they're both the same speed like that.
Not only that, but now SGI is making the Indigo series with R4000 chips
instead of R3000, upping the speed to _16_ MFLOPS ( & 85 MIPS) at 50MHz.
<climbs into asbestos suit>
Funny how Mac users always think their machines are better than anything
else at everything. :^)
<jumps behind flame-proof wall>
-Chris Wilson
cwi...@uiuc.edu
Archer Sully
*Just a note on the RISC vs CISC argument.
*
*1. MFLOPS mean ABSOLUTELY NOTHING AT ALL!!!!
*2. It takes more instructions in RISC to equal some of the instructions
* in CISC, so performance calculations (like MFLOPS and MIPS)
* get screwed up in the different instruction count values.
*3. Most important, performance GREATLY depends on the PROGRAMMER of the
* machine. Put a non-programmer on a RISC machine and a GREAT programmer
* on a Commodore 64, and I'll bet the C-64 programmer could find a way
* to make his C-64 program look faster.
Point 4. The compiler is almost more important than the CPU.
The bottom line is that the only way that you can compare two systems is
to look at their performance running a variety of appliations, such as the
SPEC suite.
I have not seen SPECmarks (of any kind) for the Mac Quadra series.
bg...@sundagger.lerc.nasa.gov
>Not only that, but now SGI is making the Indigo series with R4000 chips
>instead of R3000, upping the speed to _16_ MFLOPS ( & 85 MIPS) at 50MHz.
If you compare *current hardware*, i.e. R4000 Indigo vs Quadra 950, there's
no question that the SGI is superior. What I have my doubts about is whether
SGI (and third parties) can get enough of a variety of music or multimedia
software out to be taken seriously. I think there's only a limited time
available for this to happen; if Apple/IBM/Motorola can avoid screwing up
the PowerPC and provide good porting tools, there could be a lot of
applications, music included, in a fairly short time, after which SGI
might have a much harder sell (the installed-base effect takes over).
>
><climbs into asbestos suit>
>Funny how Mac users always think their machines are better than anything
>else at everything. :^)
><jumps behind flame-proof wall>
It's not unique to Mac users. Try defending an RS/6000 to a bunch of theoretical
chemists, most of whom have been surgically attached to SGI machines.
--
brian good
NASA Lewis Research Center
bg...@sundagger.lerc.nasa.gov
and...@cubetech.com
>Also, the WorkSpace on the Iris has to be about the best UNIX Finder
>substitute I've ever seen. Open Windows is not even close.
Yeah, it's pretty nice on the SGi's, but I think the NeXTSTEP
workspace manager is way better.
>All that said, software IS far cheaper on the Mac AND I find myself ducking
>into the console window enough to get things done that the Mac is still an
>easier system to use for non-3D, non-graphically intensive, non-UNIX sorts
>of things.
The software is expensive, but there just isn't a whole lot of it
outside of graphics and engineering. They have FrameMaker and Wingz,
and a couple more, but that's about it.
andrew
--
and...@cubetech.com | "Carpe Nocturn - I get more work done after
Andrew Loewenstern | 2 a.m. than most people do all day."
Cube Technologies, Inc. | - Simson Garfinkle
Lee Clarke
>In article <1992Nov3.1...@gn.ecn.purdue.edu> je...@gn.ecn.purdue.edu (Je
>ss M Holle) writes:
>>Also, the WorkSpace on the Iris has to be about the best UNIX Finder
>>substitute I've ever seen. Open Windows is not even close.
>
>Yeah, it's pretty nice on the SGi's, but I think the NeXTSTEP
>workspace manager is way better.
>
Personally, I like the IRIX 4.05F WorkSpace.
All one needs to know to get started with a session in the
IRIX WorkSpace is the interaction of the 3 mouse buttons and the keyboard
ALT key with WorkSpace, knowing the concept of Input Focus, and you
are off and running. The interface is very intuitive, drag-and-drop,
which makes it an excellent canditate for developing MIDI sequencer
software, editor librarians, and yes, virtual additive synthesizers.
You can issue IRIX (unix) commands in any virtual window shell and bypass
the WorkSpace concurrently with the WorkSpace running. The Console Window
is usually open at all times to at least monitor stderr messages.
The big factor is the almost extreme ease with which one can develop
windows-based apps on this machine, due in LARGE part to GL.
>>All that said, software IS far cheaper on the Mac AND I find myself ducking
>>into the console window enough to get things done that the Mac is still an
>>easier system to use for non-3D, non-graphically intensive, non-UNIX sorts
>>of things.
>
>The software is expensive, but there just isn't a whole lot of it
>outside of graphics and engineering. They have FrameMaker and Wingz,
>and a couple more, but that's about it.
>
....for the business software, yes.
There is not a lot of business software for the SGI machine. The Indigo's
are designed for the graphics, science and engineering community.
Some example products: Blue Ribbon Software Bars and Pipes, WordPerfect
5.0, Adobe Illustrator, Adobe PhotoShop (in '93), IRIS Showcase (included
with IRIX 3xx 4xx), IRIS imgworks, imgview, imgsnap, imgcopy, imgprint
(all i.w.I.),
ZIP (a very muscular IRIX text editor i.w.I.), Image Vision, Digital
Arts Rendermanager/ Render/ Animate/ Model,Ashlar-Vellum's CAD, QuickDraft,
Visix Looking Glass Professional gui, IXI X.desktop, TK Solver Plus,
Mathematica 2.0, eXclaim (Motif graphical spreadsheet, apparently Lotus
1-2-3 file compatible, with 'live-link' to FramMaker), Quicktime, and
multitrack, apanel (audio panel), cdplayer, cd play list, a gui mouse-oriented
system manager(thankyou God ! ... I come from VAX VMS...) and
drum roll please ..... IRIS Explorer....you gotta check this one out,
but Explorer discussions belong to another news group, so enough said
here.
You can get some of these products for the SUN, Mac, IBM PC, NeXT,
HP, DEC machines.
And the software may be cheaper for some platforms, but cheaper doesn't
mean faster.
Gints Klimanis
(Sean Luke) writes:
|> SGI's "expensive" apps. :-) However, the Indigo does have some
|> disadvantages compared to a NeXT. I may be wrong here, but I've been
|> told
|> that the Indigo has a DSP, but no DSP port or software access to the
|> DSP
|> itself (you can't program it). It's used only as a sound manager.
|> This
|> makes nifty programs like MetroTracks for the NeXT (a direct-to-disk
|> |> recording system)...not easy to do on the Indigo.
The idea behind the SGI Indigo DSP is to power a shared system audio
resource. This 56001 manages four stereo audio ports (port may be
either input or output). Thus multiple applications can independently
use the audio hardware with mostly transparent contention (input ports
restricted to a common input source: microphone, line or digital
(AES/EBU) and common sampling rate. Output ports limited to common
sampling rate)
To address your direct to disk recording inquiry, we have several
utilities that are shipped STANDARD with the machine.
'recordaifc'/'playaifc' are command line utilities that record/play
single files direct form disk. 'soundeditor' is a graphicial waveform
editor that records/plays single files direct from disk. The demo
application 'multitrack' plays/records multiple files from disk,
complete with mixing, two band equalization, and common echo generator.
SGI also ships 'cdman' and 'datman' (STANDARD), utilities that acuire
data over SCSI from the CDROM and DAT drives, respectively. Their gui's
are CD player front panel mockups. SGI machine options include machine
internal DAT drives that not only archive data but also read and write
audio DAT's.
WaveFrame announced TidalWave, a software only multi track
recording/editing system for the Indigo.
Sean Luke
>In rec.music.synth, and...@cubetech.com writes:
>>In article <1992Nov3.1...@gn.ecn.purdue.edu>
je...@gn.ecn.purdue.edu (Je
>>ss M Holle) writes:
>>>Also, the WorkSpace on the Iris has to be about the best UNIX Finder
>>>substitute I've ever seen. Open Windows is not even close.
>>
>>Yeah, it's pretty nice on the SGi's, but I think the NeXTSTEP
>>workspace manager is way better.
>>
>
> Personally, I like the IRIX 4.05F WorkSpace.
> All one needs to know to get started with a session in the
>IRIX WorkSpace is the interaction of the 3 mouse buttons and the keyboard
>ALT key with WorkSpace, knowing the concept of Input Focus, and you
>are off and running.
Holy cow. All you need to know on the NeXT interface is one mouse button.
The NeXT's WorkSpace Manager is very similar to the Mac Finder. Much as I
like SGI, I'd have to agree with Andrew on this one.
NeXT enthusiasts, though, should not be making too many remarks about
SGI's "expensive" apps. :-) However, the Indigo does have some
disadvantages compared to a NeXT. I may be wrong here, but I've been told
that the Indigo has a DSP, but no DSP port or software access to the DSP
itself (you can't program it). It's used only as a sound manager. This
makes nifty programs like MetroTracks for the NeXT (a direct-to-disk
recording system)...not easy to do on the Indigo.
--
Sean Luke
Brigham Young University MILK: It Comes From Cows
se...@digaudio.byu.edu
NeXTmail and nifty Mac stuff welcome
Jess M Holle
>>>Also, the WorkSpace on the Iris has to be about the best UNIX Finder
>>>substitute I've ever seen. Open Windows is not even close.
>>
>>Yeah, it's pretty nice on the SGi's, but I think the NeXTSTEP
>>workspace manager is way better.
I have played with a NeXT briefly and I found it's GUI rather confusing
compared to WorkSpace (which I have not used all that much) and the Mac
(which I have used a lot, so I may be SOMEWHAT biased here).
The GUI within programs is not the problem. Rather it was the silly three
column file executor thing (I forget the name, it's been a while) and....
other things which I forget. It has been a while, BUT I was not all that
impressed by the NeXT's interface.
Jess Holle
Nathan F. Janette
>
> The GUI within programs is not the problem. Rather it was the silly
three
> column file executor thing (I forget the name, it's been a while)
and....
> other things which I forget. It has been a while, BUT I was not all
that
> impressed by the NeXT's interface.
We have twelve NeXTs and Three SGIs. Both are great
machines, but note that we chose to sit in front of the
NeXTs all day (and night) long, and only use the SGIs for
the graphics programs written for them (and number
crunching)...and dogfighting ;-)
--
Nathan Janette
PPP link from hilbert.csb.yale.edu
Please reply to: nat...@laplace.csb.yale.edu (NeXT)
Archer Sully
*
*NeXT enthusiasts, though, should not be making too many remarks about
*SGI's "expensive" apps. :-) However, the Indigo does have some
*disadvantages compared to a NeXT. I may be wrong here, but I've been told
*that the Indigo has a DSP, but no DSP port or software access to the DSP
*itself (you can't program it). It's used only as a sound manager. This
*makes nifty programs like MetroTracks for the NeXT (a direct-to-disk
*recording system)...not easy to do on the Indigo.
*
Hmm... Then what's this Multitrack program (and TidalWave and Soundeditor)
I've been using. Coulda fooled me ;-).
As it turns out, the R3000 is no slouch at signal processing, and is also
much easier to program. Further, MIPS CPU upgrades are coming at a faster
pace than 56K upgrades.
Archer Sully <arc...@sgi.com> | Election '92: a 250 million way tie for last.
Kent L. Shephard
I've been trying to stay out of this but this is getting tired.
1. Don't even think a R3000 can compete with *ANY* 56K derivative.
Why? Because the 56K also gives you transparent data transfer both in and
out over several ports via interrupt service. This means processing samples
and never worrying about dropping samples that come in as long as your main
routine is not real long. Try a hardware loop on a R3000, I don't think so.
My MSEE project was digital audio applications hardware and software on a NeXT.
The main drawback on the Indigo is *NO* access to the 56K while on the NeXT
I can do what ever I like. System beeps and such are handled by the OS
and the 68040 and don't need the DSP.
Also a 4 track application can be written for the NeXT in about 4 hours if you
have decent skills. I know a friend of mine wrote one that does 4
tracks of stereo which means 8 mono tracks. This included panning and
volume.
The stuff I did for my MSEE was digital effects stuff. I can use my NeXT as
a digital effects processor (all digital) or an all digital EQ.
SGI machines are good for graphics but they are not the cheapest things for
digital recording. A NeXT would be cheaper.
Also note that NeXT software is more in line with PC prices than with the likes
of SGI and Sun software prices.
>
>Archer Sully <arc...@sgi.com> | Election '92: a 250 million way tie for last.
--
/* These are only my opinions. Whose else could/would take credit? */
Rainer Malzbender
>We have twelve NeXTs and Three SGIs. Both are great
>machines, but note that we chose to sit in front of the
>NeXTs all day (and night) long, and only use the SGIs for
>the graphics programs written for them (and number
>crunching)...and dogfighting ;-)
OK, ok: we have fourteen NeXT's (for the undergrads) and three SGI's (for
serious work :-), and *I* prefer to sit in front of my Indigo R4000 all day :-)
I urge you to look at an Indigo with real graphics and a real CPU (that is,
R4000 plus at least XS24 graphics) before knocking them.
Although programming access to the 56001 seems to be somewhat of a secret,
it's not really necessary given the great standard audio support provided
with the machine, and the fact that the power of the R4000 will allow many
former DSP applications to be written in C. Another outstanding aspect of
the Indigo is the speed of the GIO bus, which was 133MB/sec peak on the
33Mhz Indigo, and may be faster (?) on the 50Mhz boxes.
It's overkill for most pure audio applications (certainly I'm amazed that
anyone would port a sequencer to it), but if you're into the M-word
you can't beat it.
--
Rainer Malzbender The Flint Blood Bank is only open on Mondays, Tuesdays,
Fyzzicks Wednesdays, Thursdays, and Fridays. Saturday and
U. of Colorado, Boulder Sunday they're closed.
Gints Klimanis
(Kent L. Shephard) writes:
|> 1. Don't even think a R3000 can compete with *ANY* 56K derivative.
I agree with this statement with respect to 24x24 MAC's into double
precision accumulate. I don't agree with the following:
|> Why? Because the 56K also gives you transparent data transfer both in and
|> out over several ports via interrupt service. This means processing
|> samples
|> and never worrying about dropping samples that come in as long as
|> your main
|> routine is not real long. Try a hardware loop on a R3000, I don't
|> think so.
The Indigo uses the 56k for transparent data transfer. THAT is why the
56001 is there. We have developed a substantial wealth of code
(largely unshipped code) that processes data with no loss of samples.
Identical data acquisition caveat: don't overburden the R4000 processor
and don't overburden your 56001 with too long a main loop.
If the Indigo uses the 56001 for data acquisition, why should the
Indigo's R4000 try to do some hardware loop? The R4000 is busy blazing
64-bit data words around its primary cache at 100 MHz. Slows down some
when talking to its 1 MByte secondary cache at 64-bits and 50 MHz.
|> The main drawback on the Indigo is *NO* access to the 56K while on the NeXT
|> I can do what ever I like. System beeps and such are handled by the OS
|> and the 68040 and don't need the DSP.
This is only a drawback if you are under the erroneous impression that a
56001 is for signal processing and an R4000 is not.
The main drawback for the NeXT 56001 is that ALL of your DSP
applications are limited to 96Kx24 data space, if you expand that to the
max (I know about the NeXT from 6 months use, so pardon my comments if
the #'s are outdated). Does your home-coded sampler have a lot of
breathing room in a puny NeXT 56001 data space or would it feel less
cramped if you knew you could go to 384 MBytes on an Indigo?
Kent L. Shephard
gi...@prophet.esd.sgi.com (Gints Klimanis) writes:
>In article <3aB602G...@JUTS.ccc.amdahl.com>, kl...@cd.amdahl.com
>(Kent L. Shephard) writes:
>|> 1. Don't even think a R3000 can compete with *ANY* 56K derivative.
>
>I agree with this statement with respect to 24x24 MAC's into double
>precision accumulate. I don't agree with the following:
>
>|> Why? Because the 56K also gives you transparent data transfer both
>in and
>|> out over several ports via interrupt service. This means
>processing
>|> samples
>|> and never worrying about dropping samples that come in as long as
>|> your main
>|> routine is not real long. Try a hardware loop on a R3000, I don't
>|> think so.
>
>The Indigo uses the 56k for transparent data transfer. THAT is why
>the
>56001 is there. We have developed a substantial wealth of code
>(largely unshipped code) that processes data with no loss of samples.
>Identical data acquisition caveat: don't overburden the R4000
>processor
>and don't overburden your 56001 with too long a main loop.
I agree.
>
>If the Indigo uses the 56001 for data acquisition, why should the
>Indigo's R4000 try to do some hardware loop? The R4000 is busy blazing
>64-bit data words around its primary cache at 100 MHz. Slows down some
>when talking to its 1 MByte secondary cache at 64-bits and 50 MHz.
>
>|> The main drawback on the Indigo is *NO* access to the 56K while on the NeXT
>|> I can do what ever I like. System beeps and such are handled by the OS
>|> and the 68040 and don't need the DSP.
>
>This is only a drawback if you are under the erroneous impression that a
>56001 is for signal processing and an R4000 is not.
The 56000 still does some things faster.
>
>The main drawback for the NeXT 56001 is that ALL of your DSP
>applications are limited to 96Kx24 data space, if you expand that to the
>max (I know about the NeXT from 6 months use, so pardon my comments if
>the #'s are outdated). Does your home-coded sampler have a lot of
>breathing room in a puny NeXT 56001 data space or would it feel less
>cramped if you knew you could go to 384 MBytes on an Indigo?
Just like the Indigo in 3.0 of the NeXT os the '040 offloads a lot of
the signal
procesing stuff. The '040 does a great job at mixing and with the new
code
in the MusicKit arbitrary sample rate conversion is a snap. So my
sampler
is only limited by disk space not DSP code space.
I never said the Indigo was a bad machine it's just no the *ONLY* machine that
can do audio. The NeXT for digital audio *IS* cheaper unless you need fast
3-d graphics in color also. I don't.