Hi, Calvin,
Volume rendering is about displaying 3-dimensional fields of
data points without fitting geometric primitives to the data to
be displayed. Actually, the board is quite fast at it! If you
have a 512x512x512 voxel image, it can render it at 30 frames
per second. hey, that’s 128 MegaVoxels at 30 frames per second,
or one full gigabyte of video data at 32 bits/pixel. Volume
rendering means that you split your 3D object into voxels (that
is, “3D pixels”), and instead of using wireframes or Nurbs
surfaces to fit geometry to your image, you actually describe
the full contents of the volume as a field of data values.
You can take a look at
http://www.terarecon.com/products/volumepro_prod.html for more
information, and you can look at typical images at
http://www.terarecon.com/gallery/gallery_index.html.
The pci is a bit of a limitation, but it’s a bit like texture
mapping, where one loaded texture may go a long way. The board
memory seems to be more of a critical resource than the bus!
Alberto.
----- Original Message -----
From: “Calvin Guan”
To: “Windows System Software Devs Interest List”
Sent: Sunday, August 21, 2005 7:47 PM
Subject: Re: [ntdev] WHQL and that kind of stuff
> Alberto,
>
> Sorry I don’t have answer to your cert. question.
> It seems a pretty interesting piece of HW. May I know
> how fast it renders 3D images? What does “real-time 3D
> volume rendering” mean? IMO, PCI-X 64 bits @ 66mhz
> does not provide a decent bandwidth for read-time 3D
> rendering for PC gaming. Also, what amazes me is it
> doesn’t require external power supply. Am I correct it
> renders at a lower speed compared to thoes
> off-the-shelf 3D engines?
>
> Thanks,
> Calvin
>
> — Alberto Moreira wrote:
>
>> The hardware is a voxel engine, you can take a peek
>> at
>> http://www.terarecon.com if you want more
>> information.
>>
>> It does native 3D rendering “from the inside out”,
>> as opposed to
>> a more traditional 3D product such as ATI or NVidia
>> where
>> objects are mostly wiremeshed and then shaded. It
>> uses such
>> things as ray casting and shear-warp transforms to
>> do its job,
>> and the 3D team in charge of it has published
>> several papers at
>> past Siggraphs. A 256x256x256 image with 64 bits per
>> voxel will
>> require 2 raised to the 8*3 + 6 = one gigabyte of
>> memory, and
>> this kind of technique allows full visibility over
>> the inside of
>> the image and not only to the outside. Our typical
>> board has
>> more memory than the PC it resides in. The board
>> renders to
>> video memory, so that the resulting (2D) image can
>> be uploaded
>> or DMA’ed and displayed by the machine’s standard
>> video board,
>> or used as a texture, or be transmitted out to
>> satellite
>> locations such as a PC at a doctor’s office.
>>
>> So, this isn’t your off-the-shelf Direct 3D board!
>> What I’d like
>> to get from WHQL is a general shakeout in the area
>> of the
>> board’s interface to the rest of the machine and to
>> the
>> operating system, because the application is
>> specialized enough
>> that we feel any attempt to have a formalized WHQL
>> program for
>> it would hinder us more than it would help. I want
>> to pass the
>> driver through the standard tools - Verifier, etc. -
>> but first I
>> must get it working on the new hardware, which is
>> just now
>> coming out of the development pipeline. Also we
>> would not want
>> to go through Direct 3D, OpenGL, GDI+, or any other
>> kind of
>> graphics-intensive certification loop, because we
>> don’t support
>> any of those. The board comes with its own rendering
>> library,
>> and users call those library entry points to perform
>> volume
>> rendering functions.
>>
>> The other thing I’d like to see, and this is
>> probably a far cry
>> for someone like Microsoft, is an evaluation of the
>> OS-independence of a kernel-side component such as a
>> library. I
>> already needed this when I did OpenGL for a living,
>> but here we
>> have a core library that’s portable across operating
>> systems: we
>> run on Windows, Linux, Solaris, HPUX. The bulk of
>> the
>> kernel-side software is portable, and here our main
>> interest is
>> to make sure we don’t let any OS-dependent anything
>> creep into
>> the core. It’s in a way sort of the reverse of what
>> WHQL does
>> today. It’d be a great help to us if we had a good
>> way of making
>> sure that indeed, this OpenGL or Voxel Rendering
>> core is indeed
>> Windows-independent, yet it can be called from a
>> Windows Driver
>> and it doesn’t do anything that violates the
>> standard
>> assumptions for Windows kernel software.
>>
>> That’d be great, but again, I’m not sure Microsoft
>> would be
>> interested in pursuing such an animal.
>>
>> On a lighter note, I just got an email from the
>> Microsoft WHQL
>> people, where they clearly explain the current
>> status of
>> Longhorn certification. It’s a very nice piece of
>> paper, and I
>> have asked them permission to publish it here at
>> NTDEV. This
>> should give us a good starting point!
>>
>>
>> Alberto.
>>
>> ----- Original Message -----
>> From: “Don Burn”
>> Newsgroups: ntdev
>> To: “Windows System Software Devs Interest List”
>>
>> Sent: Sunday, August 14, 2005 9:21 AM
>> Subject: Re:[ntdev] Re:WHQL and that kind of stuff
>>
>>
>> > Alberto,
>> >
>> > May I request that as you read through this
>> stuff, you
>> > post your wish list here. As an experienced
>> developer, I
>> > think WHQL would appreciate your comments. Even
>> this post I
>> > am responding to indicates they could do better on
>> getting
>> > started.
>> >
>> > No promises but I will bug WHQL folks I’ve
>> encountered
>> > over the years to read the posts.
>> >
>> >
>> > –
>> > Don Burn (MVP, Windows DDK)
>> > Windows 2k/XP/2k3 Filesystem and Driver Consulting
>> > Remove StopSpam from the email to reply
>> >
>> >
>> >
>> > “Alberto Moreira” wrote in
>> message
>> > news:xxxxx@ntdev…
>> >> Hi, Don,
>> >>
>> >> Thanks very much for the information! Much
>> appreciated. This
>> >> is exactly the kind of thing I was looking for,
>> but there’s
>> >> so much information in the Microsoft WHQL site
>> that’s
>> >> difficult to sift down to what I needed. Thank
>> you!
>> >>
>> >> Alberto.
>> >>
>> >>
>> >> ----- Original Message -----
>> >> From: “Don Burn”
>> >> Newsgroups: ntdev
>> >> To: “Windows System Software Devs Interest List”
>> >>
>> >> Sent: Saturday, August 13, 2005 9:03 AM
>> >> Subject: Re:[ntdev] WHQL and that kind of stuff
>> >>
>> >>
>> >>> Alberto,
>> >>>
>> >>> You may be able WHQL it through the Universal
>> test program,
>> >>>
>> http://www.microsoft.com/whdc/whql/ann/ann886.mspx.
>> This is
>> >>> now a self test program, so except for a small
>> fee and
>> >>> running the tests there is little cost for the
>> effort. I
>> >>> have done work in the medical products driver
>> area, and my
>> >>> customer wanted as many certifications as
>> possible, if yours
>> >>> is similar this will probably be best.
>> >>>
>> >>> Otherwise you may want to review the two part
>> presentation
>> >>> “Device Install for Windows Longhorn” at
>> >>>
>>
> http://www.microsoft.com/whdc/driver/WDK/default.mspx
>> >>>
>> >>>
>> >>> –
>> >>> Don Burn (MVP, Windows DDK)
>> >>> Windows 2k/XP/2k3 Filesystem and Driver
>> Consulting
>> >>> Remove StopSpam from the email to reply
>> >>>
>> >>>
>> >>>
>> >>> “Alberto Moreira” wrote in
>> message
>> >>> news:xxxxx@ntdev…
>> >>>> Guys,
>> >>>>
>>
> === message truncated ===
>
>
>
>
>
>
>
> __________________________________________________________
> Find your next car at http://autos.yahoo.ca
>
> —
> Questions? First check the Kernel Driver FAQ at
> http://www.osronline.com/article.cfm?id=256
>
> You are currently subscribed to ntdev as: xxxxx@ieee.org
> To unsubscribe send a blank email to
> xxxxx@lists.osr.com