#MAX, Is it real?
34 messages in this thread
Hello All,
Can somebody tell me if the Matrox MAX is real? A month ago there
seemed to be a lot of talk on the forum about it and my local Matrox
Dealer was saying how wonderful it is. Today, there is no longer any
mention of it, and my dealer is telling me it is no good. Could
somebody enlighten me on this unicorn of a product.
1) Is it real?
2) Is it currently shipping?
3) If yes to 1 and 2, does anyone out there have one?
4) Is it really that much better quality than the PAR?
5) Any dealers in the New England area who would be willing to demo the
beast?
Thanks in advance,
Jim Grotto
James,
Let me give your questions a try.
>> 1) Is it real?<<
Yes
>> 2) Is it currently shipping?<<
Yes .. but they are final Beta versions
>> 4) Is it really that much better quality than the PAR?<<
Now this is the tough one. The sustained data thruput of the MAX is
now rated at up to 4.5 MB/sec .. which is about 10-15% better than
the PAR. The MAX is based around a high quality frame buffer
(Illuminator Pro) and thus is capable of being used for multiple
tasks .. i.e. Paint-box, Character generation, etc … where as the
PAR is a single function device. So better is in the eye of the
beholder. If you are a broadcast quality video person as well as an
animator .. I think the MAX is the best choice. If you are an
industrial animator, game 0developer, multimedia developer, etc ..
then the MAX would be "over kill" and the PAR would be a better
choice.
Hope that helps.
>> Today, there is no longer any mention of it, and my dealer is
telling me it is no good.<<
I would be interested on what basis your dealer is making this
conclusion.
Ang
DTI, An Authorized Autodesk MM and Matrox Dealer
Thanks for your reply, Angelo!
My dealer has no basis for making said claims as far as I can tell.
Your information has been most helpful. Are you running the beta
version. If so do you mind answering a few more questions!?
1) How's the setup, easy or hard?
2) Do they take the same approach with the MAX as they do with Matrox Studio,
i.e there is an approved machine list that they support?
3) Are there drivers for 3D Studio, i.e. direct render to the MAX.
James,
>> How's the setup, easy or hard?<<
Hard now .. but mainly BECAUSE it is Beta software. I can "see the
light" so to speak and the installation will be straight forward in
the final release (heck .. if it were easy .. it wouldn't be call
BETA <G>)
>> Do they take the same approach with the MAX as they do with Matrox Studio,
i.e there is an approved machine list that they support?<<
No. Any EISA system is OK. The Studio has that list because it has so many
cards .. and very few MB's have enough slots.
>> Are there drivers for 3D Studio, i.e. direct render to the MAX.<<
Yep!
Ang
I have a MAX in my machine, as we speak, and the thing is awesome. I
am getting read rates of 5.6 megabytes per second, and write rates of
4.8 megabytes per second with the recommended Seagate Barracuda 2.1
gig SCSI HD and Adaptec 2742 controller. The quality is fantastic,
but I haven't done ANY side by side comparisons. The only side by
side I'll probably do is next to an Abekas A66 (which uses NO
compression, and should be better) which I'll consider a benchmark.
Nonetheless, I'm completely impressed with the quality, and it should
hold up favorably in any of the "unscientific" tests I'm going to do
for the article I'm writing on the unit.
I have an advance beta, and there are still software bugs that need to
be worked out. But the unit works beautifully, and as Angelo said,
should be a great addition to someones advanced animation setup, and
a real benefit to anyone interested in non-linear editing, since the
board and software is also compatible with Matrox' STUDIO editing
system. The other super great feature in this system is that it's
SCSI, and therefore one can array hd's. Matrox and their software
developers are working on adding this feature as soon as possible.
This would pump throughput up to 10mb per second on my machine (and I
think that's the maximum "sustainable" throughput on EISA SCSI II
fast cards).
I have absolutley no idea why your dealer is telling you it's no good,
unless he's trying to sell you a different system.
John,
what is the approx. price of the MAX (in described configuration) ? I
have two questions:
1) Is there a PAL version ? 2) Is it possible to have dedicated slave
machine with EISA bus, relatively small amount of memory, with simple
graphic card and with MAX in 3DS network? My idea is to use this
machine only as a frame storage.
Thx PavelK
The Illuminator Pro, which is the framebuffer, is both PAL and NTSC compliant,
due to its CCIR-601 digital circutry. In the installation I am given a choice
of PAL or NTSC output (fps), so I'll assume the PAL version will be available
the same time as the NTSC. I could be wrong.
It is certainly possible to have a minimal (16megs is minimum RAM requirement)
machine, but it's nice to have the Illuminator Pro video frame buffer in your
3DS workstation, so you can put still images on the video monitor while you're
working on models and test renders and such.
The MAX, which consists of the Illuminator Pro (4meg= $3800), CAV
output board ($350), Studio Express board, and MAX software will run
between $6-7000, and I guess it depends on which model Illuminator
Pro you choose. Additional requirements are Seagate Barracuda SCSI
II hard drive (2.1 gig = @ $1900), Adaptec 2740 or 1740 series EISA
controllers (@ $390), and an EISA motherboard with 16megs ram
minimum.
John,
One thing to note about the Seagate Barracuda drives is that access
seems to slow down as they fill up.
Kevin Krell – Computer Support Associates
That may have to do with using the slower, inner tracks of the HD as it fills
up, Kevin.
-Alan
Alan,
The Micropolis drives don't slow down, though.
Kevin Krell – Computer Support Associates
Right. Hmmm…I guess I didn't word that properly, Kevin. What I meant
was, because the inner tracks are closer to the hub of the hard
drive, they actually travel slower under the heads (compared to the
outer tracks), making for a slight decrease in performance. This is
especially noticable in the newer, AV-class drives. As the drive
fills up (from the outside in) the performace hit can sometimes be
noticed. At least this was how it was explained to me by a hard drive
designer-type person.
Alan,
I retract my statement. Apparently, the performance hit will
occur on any drive – it just differs as to when it becomes
noticeable. "Never mind…"
Evidently the problem has only recently been noticed more often with the
AV-class stuff.
-Alan (has "lost it" a few times…<g>)
I'm not exactly sure how the MAX software gets such great througput, but I know
it takes over the Adaptec controller and does its own thing. The test software
that gave me the results writes a file over the entire 2.1 gigs of the hard
drive, so I think it mush take a filled drive into consideration when giving
results and setting q-factor parameters. But I don't really know for sure.
John Tissavary (LUNA cie)
Hey John, can you manually choose the amount of hard drive
space-per-frame on the MAX system like we do on the PAR? Or is it
"pre-ordained" for you?
I haven't explored all the switches, and I know some features in the
GUI aren't activated yet, but you can set a "limiter" in the max.ini
file, which will put a ceiling on the largest frame size. The
q-factoring is dynamic (as is the PAR, am I right?), so frames that
are less than the ceiling will be the same as before, but frames over
the ceiling will be further compressed.
John Tissavary (LUNA cie)
What sort of performance do you obtain off of the A/V HD? (If
you don't have the A/V HD, ignore me)… i.e. on what kind of bus do
you get what kind of performance…
Yes, John, Q-factors change dynamically from frame to frame. But the
actual amount of compression on any given frame depends greatly on
the Block Limit setting. The higher BL you can get away with (depends
on the performance characteristics of the hard drive), the less
compression, or higher Q-factor will be used for any given frame.
This allows the PAR user to actually select the *amount of hard drive
space* used per frame for a given animation, optimizing image quality
for the particular hard drive and frame characteristics being used at
the time. I thought that this was a pretty slick feature of the DPS
system, and wondered if the MAX had the same feature. Maybe the MAX's
HD control/compression method is so good, you don't need to manually
set it for every animation? Of course, the quality had better be
pretty lossless if this is the case, IMHO.
Hope you are having fun, John!
-Alan (PARticular about features…)
There is a program that one runs, called Profiler, that determines the
highest possible throughput on the entire disk, and sets the q factor
limitations accordingly. Additionaly, one can edit these values
manually in the max.ini file, although I don't know enough about the
number values to do that effectively myself. I have gottem good
results just trusting Profiler.
Sounds interesting, John. Thanks for the info.
-Alan
Please keep us all posted on your comparisons. My company is
considering the MAX, but we are no way video output experts. I have
had the opportunity to dump about 6 sec of 3D animation to plain VHS
tape via a MAX on loan to my VAR – I thought it looked great, and I
couldn't see any compression artifacts until he pointed them out to
me. (Our consumer-level medium is VHS tapes). We do have the option
of going the Exabyte/Abekas route instead (we already have an
Exabyte), but the capacity to do truecolor realtime tests and demo
tapes ourselves is really tempting, (then renting a beta deck to do
masters).
We are not likely to do the character-generation and post-prod work ourselves,
i.e. no plans to get involved in Matrox Studio, so maybe the MAX is overkill
for us.(?).
The particular MAX system I tried, which was the system being shipped
around to different shows and conferences, was appallingly
crash-prone. Any comments on this? The problem was aggravated after
it had been working a couple of hours, and was exacerbated by any
kind of editing efforts – have you had such problems on your system?
Could this be due to all the shipping stresses, or the newest
software version which was loaded on just before shipping (about a
week and a half ago).
Since you are actually working with it, any feedback will be
appreciated – it will be a BIG outlay for us, if we get it, and I'd
sure hate to make such an expensive mistake.
Thanks.. Gayle G de N, SAVANTES
After some intial problems with installation, I have managed to record
2700 frames (and play them back in non-linear fashion!) without
problems. There are still bugs in the software, but it's still in
beta, so there's nothing surprising there. The MAX is now running
99% solid – I had a crash once, when playing back the 2700 frames
non-linear, but several other tries were successful.
My initial problems were due to what seemed like hardware problems
with the MAX itself, but turned out to be bugs in the software
drivers. Matrox, et al, are working VERY hard to rectify all
reported bugs, and have modemed me new betas within 24 hours.
I beleive that once the software is stable enough for commercial
release, the installation should be fairly simple – comparable to
getting 3D Studio running on a Lantastic network (simple is in the
eye of the beholder <g>).
So far I'm floored by the output, but I haven't been near an Abekas in
a little while <g>.
John Tissavary (LUNA cie)
<< comparable to getting 3D Studio running on a Lantastic network >>
I remember your November <BG>
jonas[adesk]
Jonas or John, sorry if I am little slow, but I missed the discussion
about Lantastic/3DS netrender in november. We have some problems
ourselves getting it to work reliable. Machines stop rendering
without any clear reason. Is there a simple source for answers ?
Regards, Fremer Huijgen (Studio PC)
It depends what is going on at the time of the render failures, but
with Lantastic and 3DS net rendering, when more than one machine
tries to access the same file and it does not have "read-only"
attributes, all but one machine will fail. My solution is to shell
to dos before a net render, and switch all .prj, maps, and other
necessary files to "read-only".
If that doesn't sound like the problem, give me more details and I'll
take another crach at it.
John,
When we talked last, you had gotten 4.5MB/sec out. What's your latest numbers?
Ang
John, your hunch is partly right. It seems part of the problem, but
even we set all files to Read-only we experience machines 'falling
off' during long sessions. Recently I thought it could be
time-related. The clocks (dos-clocks I mean, not the CPU-clock) of
the machines tend to drift apart, so maybe the message passing fails
after some time. Could this be ?
Fremer,
<< lantastic and november >>
Resulted in a paper called 3dsnet.txt and found in the 3d Studio
section of the libs.
Also, with many networks, one goal is to minimize network traffic. So
keep your Mat Libs, IPAS', rendered images, all local to the slave.
Another goal is synchronization of the slaves and some individuals
will SYNC the net automatically many times over the day. John refered
to a read only issue. Setting the read only flag on shared files
will improve problems in network failure caused from Net-Timeouts
waiting on a file that is opened for read/write privaledges when open
for read will do.
Thanks Jonas. As a matter of fact my last hunch was indeed that we
could have time related problems between the different dos-clocks.
Will read the 3dsnet.txt and hope it contains the answers.
That's really impressive – 2700 frames non-linear! Our test was very gentle by
comparison. I find your results really encouraging. The "output end" of
SAVANTES just had a look at the test tape I did, and was delighted. We'll
probably go for it – but we'll wait as long as possible to make the purchase,
hoping the software will "stabilize".
BTW – my Lantastic works great with 3D Studio – (of course I'm not the one who
set it up – maybe a lot of anguish went into the working network I got <g>!).
I wouldn't say it's fast when VPost is pulling source files over the network –
but within 3D Studio it has been crash-free (knock on six kinds of wood). The
only thing that reliably crashes the works is if I do any graphics-intense
Windoze apps while the server is working hard in 3D Studio. This is annoying,
because I like to tweak texture maps over the net while I'm working in the
Materials Editor. I think this is a Windoze problem, more than the Lantastic,
though I don't have any good reason for thinking so (except that the net
usually works, and Windows very often does not!) Though I really like Fractal
Painter for textures, I end up not using it interactively because of this
problem – I use HiResQFX (which runs in 32-bit protected mode in DOS, and is
blazingly fast – but not nearly as much fun).
Keep the MAX thread going!
Gayle G de N, SAVANTES
You may want to ensure that your windows LAN driver is installed. I had
similar problems at first, but then read the Lantastic manual (imagine that…
<g>) and found out there's a little statment you put into the system.ini that
loads a windoze LAN driver. I haven't crashed the net since, even with 32 meg
photostyler and Fractal images.
The Max is shaping up to be really great. I've gotten more done in two days
than I did in two weeks of single framing. Can you imagine organizing 24
scenes, 2700 frames for single frame recording? Took me less than a day to do
with the MAX.
John Tissavary (LUNA cie)
5/27/94
What! You read the Lantastic manual?! I will try your suggestion (uh, your
specific suggestion – not RTwholeM <g>).
BTW, I downloaded the 3DSNET.TXT file, and realize that I probably haven't had
many problems with Lantastic because I'm really not pushing it (also, I only
have 2 systems connected), and for the most part I had been distributing the
projects so that the slower machine got small, sequential easy ones, and wrote
to its own HD. In fact, I would just do batch renders on that machine, except I
really like having the detailed report on the render that controlling it over
the network provides. The first time I tried using the network for a big VPost
project, with large source files for compositing drawn over the network, and
both systems writing to a MO, it was so slow I thought it HAD crashed.
5/31/94 – I just checked my system.ini files for the additions recommended by
Lantastic – interestingly, all the lines were there on one system, and none on
the other; I remedied that, and will see how it goes. I'll really be delighted
if I can use Fractal again without crashes – so thanks much for the suggestion.
Gayle G de N, SAVANTES
Thaks, John,
I think you might be correct about my dealer. I am trying to make a smart
choice. He is trying to make a quick sale!
Jim Grotto
Hi James:
I just got back from Comdex in Atlanta and had hoped to see the MAX up
and running, better than it was in December at the Video Expo.
So, it is real and has changed somewhat since December. Matrox has
improved the user interface and has now defined the EISA SCSI
controller card and hard drive that is approved for this
application. The Matrox rep says it is still not available due to
several bugs that need to be worked out but it has shipped to several
"alpha/ beta" test sites. I asked when can we buy and he said about
mid summer. The demo of animation was still the same rollercoaster
animation I saw in December and it still was crashing the system
after about 15 seconds of playback. The rep had to reboot the
machine to re launch the software. He said that was still one of the
bug fixes that is currently being tested by beta testers. I liked
the file interface and the MAX had pretty good quality. The aliasing
on diagonal edges looked about the same as the PAR and the
pixelization in the busy areas of the picture looked the same as the
PAR, so, given the proposed better file transfer rate of the MAX I
would say that this buggy version that was being demoed was still not
producing the far superior quality I have been hoping to see as
predicted by all the crystal ball gazers on this product during the
last few months. I understand that John Tissavary has one of these
beta/alpha systems under test. Maybe he could further enlighten us
as to how it works.