Posted by
CmdrTaco
on from the hardware-to-lust-after dept.
BlackHat linked us to IBMs
Deep View, a
research system for rendering and other advanced applications (Q3A). The
PC is 8 Linux boxes in a rack, which is needed to generate the content for the
T221 display which operates at 3840x2400.
Re:This is a software solution
by
ObviousGuy
·
· Score: 2, Insightful
I knew I misphrased that as soon as I hit submit. This is an advancement in algorithms (hardware-based in this case), rather than any special functionality of an OS.
-- I have been pwned because my/. password was too easy to guess.
Bumps the Mig-19 from my list
by
anticypher
·
· Score: 4, Funny
9.2 megaPixel, up to 56 Hz refresh in a 22 inch LCD screen. Want!
This is now very high on my luxury wish list. When is the next IT bubble scheduled to happen again? I have to start my plan to get rich on other people's stupidity and greed so I can afford a system like this.
the AC
-- Hemos is like...sci-fi fans;he thinks technology is cool, but he hasn't bothered to understand the science it's based on
Re:Bumps the Mig-19 from my list
by
be-fan
·
· Score: 2
Well, the referesh rate on an LCD doesn't matter that much. 56 HZ doesn't cause any flicker. And its low enough ( 25 ms response time) that you probably wouldn't see any ghosting in games.
-- A deep unwavering belief is a sure sign you're missing something...
Re:Bumps the Mig-19 from my list
by
be-fan
·
· Score: 2
We'll, its more like 60 fps average, though. For most people as long as it doesn't go below 40-50 fps minimum it looks fine.
-- A deep unwavering belief is a sure sign you're missing something...
Quoth the article: since no single graphics adapter has the necessary horsepower and bandwidth to feed a 9.2 million pixel display (at 41Hz using 24 bits per pixel)
Hmm, doing the math: 3840*2400 pixels = 9216000 Pixels per Frame 9216000*3 (3=24/8) = 27648000 Bytes per Frame 27648000*8 = 221184000 Bits Per Frame 221184000 (bpf) * 41 (fps) = 9068544000 Bits per Second 9068544000/1024 = 8856000 KiloBits per Second (approx) 8856000/1024 = 8648 MegaBits per Second (approx) 8648/1024 = Just over 8 GigaBits per Second
Now, with newer DX9-type graphics Adapters, and AGP 8x, we can do about 2.5 to 3 Gigabits per second just now (over the AGP Bus, haven't calculated Actual Display bitrates!)... Applying Moore's law, (theorum, whatever!), we can safely say that this kind of horsepower will be common in a single, average, desktop PC inside of two to three years.
Sure, this may be a boost to Hollywood today - but soon enough, it will be pretty commonplace technology. (Though I'm betting the most expensive item in that bunch of kit is the actual LCD Display, not the kit driving it!).
-- Disclaimer: I meant what I thought, not what I wrote! What? You can't read my Mind? Oh dear!
Re:Bandwidth...
by
be-fan
·
· Score: 5, Informative
Umm, just over 8 gigabits per second is nothing. That's about 1 GB/sec, and an 8x AGP (2 GB/sec) graphics adapter (Radeon 9700) would have no trouble handling that data rate. Besides, you don't have to pump *all* of that data over the AGP bus. You send only display lists and textures and whatnot over the AGP bus. The local graphics card (where the actual data rate would hit near 1 GB/sec) has (on a Radeon 9700) about 20 GB/sec of bandwidth.
-- A deep unwavering belief is a sure sign you're missing something...
This used to be almost exclusively SGI's territory.
A few years ago I attended a presentation at NCSA where Larry Smarr was talking about their plans for a similar display, driven by about $1M of SGI boxes. I think they wanted to call it "The Great Wall of Power".
Some future PlayStation "n" will do this in your living room.
Isn't Moore's Law great?
Which leads me to ask, have we ever had a/. poll on your favourite law? -Moore's -Murphy's -Amdahl's -Newton's - Hooke's -Boyle's -whatever?
Great game engine
by
johnlcallaway
·
· Score: 5, Interesting
The human eye only resolves 12 to 15 frames per second
Where did you get that from?? It's easily proved wrong by the fact that you can see the flicker of a 60Hz monitor, but even then, if that's all you could see it would be impossible to play sports (such as hitting a baseball).
The whole "what is the frame rate of the eye" question is an incredibly complex one that I won't try and do here (yet again). Suffice it to say that it's not as simple as "frame rate".
-- Sometimes it's best to just let stupid people be stupid.
your brain will quite happily fill in the missing detail about an object's position, be it a Baseball or whatever, quite unconciously - you don't have to see it to know it's there and where it's going!
Spoken like someone who doesn't play sports (no offense). Anticipation is certainly an important part, but only a minor part. You will NOT be able to hit a 80-90 MPH curve ball by just watching the angle coming from the pitcher's hand. Think about the margin for error in hitting a baseball, and then think about the angle you are watching the baseball come in at, which is basically straight at you.
Or heck, an easier experiment is to have someone toss a ball to you and close your eyes at the halfway point when it reaches the top of the arc. It will NOT be easy to catch, but by your theory is should be since you have data from half the travel of the ball.
This really shouldn't be surprising. Which would be easier to design... a device that tracks a ball and then tries to predict where a grabber should be to catch it, or a device that watches the flight of the ball and continuously updates the grabber position based on where it sees the ball going? Again, anticipation is certainly a part of it since you want a rought approximation of where the ball is going to be, but it's not nearly enough data to be accurate.
-- Sometimes it's best to just let stupid people be stupid.
I've gotta call bullshit there. The only case where lighting would have any effect on the perception of a CRT would be its intensity, and even then its effect would be indirect.
Your eyes are much more sensitive to flicker in the periphery than looking dead on. If the room is very bright, your eyes will be less dialated, and you'll be less sensitive to the flicker of your monitor.
Artificial light does have it's own flicker component, but that won't interfere with a crt because it doesn't depend on the reflection of that light for its operation. Now if you take an HP48 calculator, you will probably notice some flicker in rooms lit mostly with flourescent lights. The refresh on flourescents (in the US anyway) are close to the refresh of the reflective lcd on the HP48, hence the banding.
Another big factor on flicker is the rate of decay of the phospher elements in your monitor. The slower the glow decay, the less likley you are to see a flicker--the pixel is still glowing from the last time it was hit, when it is struck again. The longer this decay, the lower refresh rate you can get away with from a flicker point of view.
However, now you suffer from smearing or stuttering (sometimes called ghosting). The optimal setup would be a phospher coating with nearly infinite decay rate, operating with an infinite refresh rate.
Television, here in the US is refreshed at something like 30Hz (non hdtv). The reasons you don't see the flicker are: 1) slow decay rate of the phospher. 2) you are usually 5 or 6 feet away from a tv when you are watching--so it isn't in your peripheral vision. 3) While big screen tv's are getting more common, most people are still below the 36 inch mark, which also means it is mostly in your non-peripheral vision.
Try this: Go up to like a 13 inch tv or something small like that, turn it on to some show that has a lot of white to it. Stand about 1 foot or so away, and look just above the TV. I guarantee that you will see flicker. Some people are more sensitve to flicker than others, and it will depend a little on the TV, but at 30hz, I imagine everyone in the world can see it.
LCD--I think all of the consumer LCDs out today suck as far as pixel decay. I don't know the reason, capacitance maybe, but they suck. Much slower decay than CRTs. On many LCDs today, you still can't tell if you have "mouse trails" turned on or off (in ms windows). So that is why you don't see flicker as much at such low refresh rates on LCDs. There may be other reasons too... I don't know.
Hey listen, all you said was that you were "partially sighted", which doesn't mean squat, and that your vision was 20/120. Maybe you meant something else, but 20/120 in the US means you can almost get by without glasses at all.
"You can say what you want, and you can Mod this post all the way to oblivion, but you cannot change how anyone else sees the world"...nor can you enlighten a fool.
Now we have anonymous fools. My vision is 20/300 in my right eye, 20/400 in my left, and with glasses I see just fine. So quit this cry-baby shit about 20/120 vision being some kind of handicap.
On a side note, I think the tendency here in america is to measure vision in diopters.
Re:This is a software solution
by
You're+All+Wrong
·
· Score: 2, Interesting
No need to retract, Linux isn't the most important part of the set-up at all.
I did't see anywhere where it said what processor it's using though. A 866MHz what?
Chips at that speed include PIII and Alpha 21264, but not as far as I know a Power or PPC speed (or HPPA, MIPS, or Sparc).
I know Alphas are popular in render-farms, but have gone 'out of fashion' now. Are IBM _embarassed_ by their choice of processor?
YAWIAR.
--
Your head of state is a corrupt weasel, I hope you're happy.
The death of the zoom tool
by
HawaiianMayan
·
· Score: 5, Interesting
IBM brought one of these screens by Alias|Wavefront to show. The image detail is unbelievable.
In fact, you don't need a zoom tool on your paint program anymore,.. you just need a real magnifying glass sitting next to the monitor (IBM brought one), because it's showing much more detail than you can really see!
One thing it shows, though, is the need for vector-bases scalable interfaces... the default Windows UI was so tiny on that screen it was really hard to use!
Re:The death of the zoom tool
by
zapfie
·
· Score: 2
That's actually something I've been thinking about.. the idea I finally came up with after brainstorming it a bit is that you would need two coordinate systems, one in pixels, and one in some other unit system of your choosing. You could then specify a button as 5 units by 1 unit, and set the screen to be 64 units by 48 units or something. This would allow you to a) change the size of the workspace independent of the screen resolution and b) change the screen resolution independant of the workspace size. Of course, you would probably need some safeguard to make sure you didn't have a workspace that was much larger than your current resolution could handle, which might result in unreadable text, etc... Back to your original point though, I definately second the notion it's high time for some vector-based interfaces.
Ummm and Sony have announced that IBM and Toshiba will be joining up to develop the architecture and processors for the Playstation 3.
The odds on these two pieces of work not being related have to be pretty slim. Its a pretty clear gameplan, XBOX2 is a "Windows Home Gateway", PS3 is a "Multimedia Home Gateway" that happens to be running linux.
-- An Eye for an Eye will make the whole world blind - Gandhi
This is the stupidest thing I've ever heard The odds on these two pieces of work not being related have to be pretty slim. Its a pretty clear gameplan, XBOX2 is a "Windows Home Gateway", PS3 is a "Multimedia Home Gateway" that happens to be running linux.
You have absolutely no proof to back up these claims. Do you know HOW much research IBM does? Here is a list of research products [ibm.com]. Thats about 400 different projects that IBM is currently working on. Given the amount of projects I'de say there is a greater probability that they're NOT related. Except that they both involve parallel processing.
-- A rabbit in the hand is worth 4 in the cage
Which is why you use an LCD...
by
MosesJones
·
· Score: 4, Informative
Films are 24Hz, but you don't worry about that, and using an LCD screen means that you don't get the blurring and flashing of a normal monitor.
-- An Eye for an Eye will make the whole world blind - Gandhi
Re:Which is why you use an LCD...
by
donglekey
·
· Score: 2
Actually film runs at 24 frames a second, and even though this is enough for the illusion of motion, persistence of vision does not stretch that far. Projectors in movie theatres open and shut twice every frame, so standard theatre film movies are shown at 48Hz
... I know! Lets have a press release about our high resolution graphics products, and put a high resolution picture in with the article!
Oh, and while we're at it, lets make it a 256 color gif!
-- Karma: SELECT `karma` FROM `users` WHERE `userid`=138474;
Re:56 Hz? (LCD)
by
qnonsense
·
· Score: 2, Informative
It's an LCD. It's not like a monitor that flickers. The glow is fully solid, it's just to color of the pixels that refresh at 56Hz. That's plenty fast when the whole damn thing ain't flashing at you.
-- There comes a time in every man's life when he must say, "No mother! I do not want any more Jell-O!"
One thing puzzles me about the images that were shown from popular games (Flight Simulator, Quake III, etc.): Why did the graphics look so fuzzy at that ultra-high resolution? The site mentioned that texture-mapping on Quake III isn't optimized for that resolution, but my question is:
Is the point that it runs in real time at that resolution (even though it looks mediocre) or is the point that it supposedly looks great at that resolution? If it's about real-time high-res, then that makes sense to me, but practically speaking who cares what the resolution is if the image quality doesn't improve with higher resolution as seems to be the case with Quake III and Flight Simulator.
um, you can't expect the actual textures to look better just because they are shown with better resolution (in fact, quite the opposite). the "point" is that you can run it in real time with that resolution, and make software that has high enough resolution textures and looks really good. It's kinda like bumping your monitor resolution to 1600x1200 and expecting old DOS games to look better as the result.
-- sic transit gloria mundi
The original poster is not a troll, here's why.
by
wackybrit
·
· Score: 2
He has a point. A dual headed GeForce 4 can drive two displays at 1600x1200 without problem, which comes out to 3200x1200. Okay, that's not quite 3600x2400, but we're not talking miles off. Why does it require 8 Linux machines just to rustle up the power to display images on a screen of that resolution?
I could understand if the Linux boxes were running a powerful simulation or something, but surely we can get devices of that resolution running on a single PC with some pretty intense hardware.
Or, is GeForce 4 et al really on the cutting edge? What do the people with millions of dollars use? Do they have to start using multiple machines like IBM? Sounds unlikely to me. What about the military? Surely someone is one step ahead of the latest consumer technology?
Why, do you have performance benchmarks comparing the experimental, non production DeepView with the commercially available InfinitePerformance visualization system?
Chromium, the software that makes it all possible
by
thatguymike
·
· Score: 2, Interesting
Chromium is a project at Stanford. There was a paper published at this years SIGGRAPH which discusses how the T221 is driven by a cluster and the SGE, as well as other applications including a parallel volume renderer. (http://graphics.stanford.edu/papers/cr/)
Chromium is an open source project and you can get it from http://chromium.sourceforge.net. Chromium is designed to enable people to harness the power of a "graphics cluster" and/or use multiple displays. You don't have to buy a T221 and an SGE to render Quake at high resolution, you can use multiple monitors/projectors instead.
-Mike
Re:Sounds Like
by
YeeHarr
·
· Score: 2, Interesting
I have seen this display driven by one Octane2.
http://www.sgi.com/workstations/octane2/
Not one rack of 8 PCs and one half rack of some graphics engine.
It was drawing the full display at about 30fps.
It was as easy to use as any other workstation rather than the 'interesting' mix of 8 pcs in a rack and some other half rack of graphics stuff.
The hardest thing was to read the tiny little fonts on the screen (the display is 200dpi IIRC) - you need a magnifying glass (or of course you could increase the font size).
The Octane2 can do this because you can install two V12 graphics engines each with a dual channel adapter.
http://www.sgi.com/workstations/octane2/dual_cha nn el.html
SGI software stiches the cards together transparently.
It is a beautiful display.
The amuzing/annoying thing about these sorts of announcements is that customers have been using SGI stuff to do this for the last 8 years or so.
If you were on the leading edge of this kind of work would you wait that long for some kludged together solution which might work if you have enough Duct tape to stick it together?
Or would you pay the extra cash for a solution that works and gives you a huge jump on your competition.
The sort of software layer that can be used to make these four channels (two channels from each graphics card) into one display is stuff like:
http://www.sgi.com/software/multipipe/sdk/
Oh yeah - and re your 'every table comment'. The Octane2 fits on one table - it doesn't surround the table like the IBM stuff.
Since Chromium allows applications to demonstrate pretty much linear scalability as nodes are added, I would guess that they got interactive frame rates. At least, that's what I've seen with Chromium displaying to one of those displays at Stanford.
SGI did this two years ago, as a *product*
by
green+pizza
·
· Score: 2
SGI did the graphics cluster thing over two years ago and even released it as a product. Very, very similar to the IBM. They even ported several of their programming APIs and SDKs from InfiniteReality to the Linux Graphics Cluster. Not many were sold, however. Heck, Slashdot didn't even cover it. It was still pretty neat to see multiple spanned monitors and even composited high res projectors driven by a half rack of Linux PCs. Many of the demos were actually ported from SGI's big iron Onyx machines and worked just as well on the cluster. The basic setup was a stack of rackmount Linux boxes using nVidia AGP cards and custom PCI cards daisy chained together to provide sync for glSwapBuffers among other things. Also availble were gigE and Myrinet for networking the machines with something better than 100BT. A compositor (similar to what's used in InfinitePerformance) was also available.
NCSA has been making good progress on developing this. The million bucks of SGI hardware has of course been replaced by a rack of Linux PC's. Instructions are actually on line. It's not trivial to build, though. And using LCD projecters does have real downsides. If you buy a dozen identical projectors, they won't have the same brightness and color saturation. Actually, they aren't even consistent from edge to edge. So your display isn't perfect, and you can definitly see the tile edges. Not to mention the fun of building a usable support structure which lets you get all the alignments right. If you don't need it to take up an entire wall, I think the IBM T221 display is a cheaper way to get super high-res output. But of course your high-res Quake won't be lifesize either.:-)
Re:This is a software solution
by
DavidRavenMoon
·
· Score: 2
I did't see anywhere where it said what processor it's using though. A 866MHz what?
Most of IBMs big machines seem to run on PPC chips of one form or another, and 866 is a common speed for G3 CPUs.
Deep Blue ran on the old PPC604e's!
I would think they are using IBM processors.
-- --
if it was so, it might be; and if it were so, it would be; but as it isn't, it ain't. That's logic - Lewis Carrol
IA-64 Itanium perhaps? However, looking at Intel's specs I don't see 866 listed as an available clock speed. They list the original Itanium at 733 and 800 MHz, and Itanium 2 at 900 MHz and 1 GHz.
I would have thought that IBM would plug their own hardware whenever possible -- the T221 display is certainly phenomenal, and they provide a link so you could buy one of those... It has me wondering how I can come up with $8400 to get one. (Heck, when those things came out, they were $20,000! Ah, progress.)
So, this leaves us to wonder... no mention of processors, "low" clock speed -- compared to what we're used to seeing -- something new from AMD? IBM pissed off at Intel? Some new massively parallel top secret silicon from IBM?
Watch... it'll turn out to be Pentium IIIs -- they call 'em "workstations," so they might have recovered them from some other project. (Or all the engineers got new workstations and wondered what cool project they could do with their old ones... Q3A at 3840x2400? What the heck!) What's the limiting factor in this case, processor power or network bandwidth?
--
"...America's great minds of today, teaching America's great minds of tomorrow. Poor bastards." -- A Beautiful Min
Re:Intel again? Why not Power4?
by
rusty0101
·
· Score: 2
I would venture to guess because the Power4 processor is not optimized, nor designed to be used in a workstation environment. I would also suspect that it is not designed to be used as a graphics processor.
Power chips are not PowerPC chips. If you are refering to PowerPC Generation 4 processors, your message is a missleading question. Additionally, I don't believe that the PPCg4 is quite up to an x86 processor (off the shelf speeds) yet. Don't get me wrong, for the work I do, a G4 Mac would probably be sufficient, but if you are talking raw graphics processing, that's a different market.
-Rusty
-- You never know...
Re:This is a software solution
by
DavidRavenMoon
·
· Score: 2
That would have been my first guess too, but in a scan down www.spec.org, and a google for ``866MHz PPC'' I found _nothing_ that matched at all.
Apple used to sell an 867MHz G4 from July 01 - Jan 02.
That's a Mototola part of course, but I'm sure IBM has a similar PPC CPU.
-- --
if it was so, it might be; and if it were so, it would be; but as it isn't, it ain't. That's logic - Lewis Carrol
Slashdot Mirror
An article about the methods IBM used to cluster eight dual processor Linux workstations to provide the necessary graphics power.
Movies of Deep View in action.
I knew I misphrased that as soon as I hit submit. This is an advancement in algorithms (hardware-based in this case), rather than any special functionality of an OS.
I have been pwned because my
9.2 megaPixel, up to 56 Hz refresh in a 22 inch LCD screen. Want!
This is now very high on my luxury wish list. When is the next IT bubble scheduled to happen again? I have to start my plan to get rich on other people's stupidity and greed so I can afford a system like this.
the AC
Hemos is like...sci-fi fans;he thinks technology is cool, but he hasn't bothered to understand the science it's based on
Quoth the article:
since no single graphics adapter has the necessary horsepower and bandwidth to feed a 9.2 million pixel display (at 41Hz using 24 bits per pixel)
Hmm, doing the math:
3840*2400 pixels = 9216000 Pixels per Frame
9216000*3 (3=24/8) = 27648000 Bytes per Frame
27648000*8 = 221184000 Bits Per Frame
221184000 (bpf) * 41 (fps) = 9068544000 Bits per Second
9068544000/1024 = 8856000 KiloBits per Second (approx)
8856000/1024 = 8648 MegaBits per Second (approx)
8648/1024 = Just over 8 GigaBits per Second
Now, with newer DX9-type graphics Adapters, and AGP 8x, we can do about 2.5 to 3 Gigabits per second just now (over the AGP Bus, haven't calculated Actual Display bitrates!)... Applying Moore's law, (theorum, whatever!), we can safely say that this kind of horsepower will be common in a single, average, desktop PC inside of two to three years.
Sure, this may be a boost to Hollywood today - but soon enough, it will be pretty commonplace technology. (Though I'm betting the most expensive item in that bunch of kit is the actual LCD Display, not the kit driving it!).
Disclaimer: I meant what I thought, not what I wrote! What? You can't read my Mind? Oh dear!
This used to be almost exclusively SGI's territory.
/. poll on your favourite law?
- Hooke's
A few years ago I attended a presentation at NCSA where Larry Smarr was talking about their plans for a similar display, driven by about $1M of SGI boxes. I think they wanted to call it "The Great Wall of Power".
Some future PlayStation "n" will do this in your living room.
Isn't Moore's Law great?
Which leads me to ask, have we ever had a
-Moore's
-Murphy's
-Amdahl's
-Newton's
-Boyle's
-whatever?
I rarely read replies, it's my opinion and if you thought about your opinion a little more, I'm OK with that.
Why not just turn your monitor on and off as fast as you can? You'll save a lot of money.
-- http://frobnosticate.com
41 hz ... ouch! That thing must look like a strobe light.
Skiers and Riders -- http://www.snowjournal.com
No need to retract, Linux isn't the most important part of the set-up at all.
I did't see anywhere where it said what processor it's using though. A 866MHz what?
Chips at that speed include PIII and Alpha 21264, but not as far as I know a Power or PPC speed (or HPPA, MIPS, or Sparc).
I know Alphas are popular in render-farms, but have gone 'out of fashion' now. Are IBM _embarassed_ by their choice of processor?
YAWIAR.
Your head of state is a corrupt weasel, I hope you're happy.
IBM brought one of these screens by Alias|Wavefront to show. The image detail is unbelievable.
In fact, you don't need a zoom tool on your paint program anymore,.. you just need a real magnifying glass sitting next to the monitor (IBM brought one), because it's showing much more detail than you can really see!
One thing it shows, though, is the need for vector-bases scalable interfaces... the default Windows UI was so tiny on that screen it was really hard to use!
Ummm and Sony have announced that IBM and Toshiba will be joining up to develop the architecture and processors for the Playstation 3.
The odds on these two pieces of work not being related have to be pretty slim. Its a pretty clear gameplan, XBOX2 is a "Windows Home Gateway", PS3 is a "Multimedia Home Gateway" that happens to be running linux.
An Eye for an Eye will make the whole world blind - Gandhi
Films are 24Hz, but you don't worry about that, and using an LCD screen means that you don't get the blurring and flashing of a normal monitor.
An Eye for an Eye will make the whole world blind - Gandhi
... I know! Lets have a press release about our high resolution graphics products, and put a high resolution picture in with the article!
Oh, and while we're at it, lets make it a 256 color gif!
Karma: SELECT `karma` FROM `users` WHERE `userid`=138474;
It's an LCD. It's not like a monitor that flickers. The glow is fully solid, it's just to color of the pixels that refresh at 56Hz. That's plenty fast when the whole damn thing ain't flashing at you.
There comes a time in every man's life when he must say, "No mother! I do not want any more Jell-O!"
Deep throught was IBM's predecessor to Deep Blue.i ne/33_f older/33_articles/33_kasparovibm.html
http://www.azer.com/aiweb/categories/magaz
(* Deep throught was IBM's predecessor to Deep Blue *)
It is rumored that the Bush admin is using IBM computers to plan an Iraqi invasion. It will be called "deep shit" (for which side is unclear).
Table-ized A.I.
One thing puzzles me about the images that were shown from popular games (Flight Simulator, Quake III, etc.): Why did the graphics look so fuzzy at that ultra-high resolution? The site mentioned that texture-mapping on Quake III isn't optimized for that resolution, but my question is:
Is the point that it runs in real time at that resolution (even though it looks mediocre) or is the point that it supposedly looks great at that resolution? If it's about real-time high-res, then that makes sense to me, but practically speaking who cares what the resolution is if the image quality doesn't improve with higher resolution as seems to be the case with Quake III and Flight Simulator.
Amazing magic tricks
He has a point. A dual headed GeForce 4 can drive two displays at 1600x1200 without problem, which comes out to 3200x1200. Okay, that's not quite 3600x2400, but we're not talking miles off. Why does it require 8 Linux machines just to rustle up the power to display images on a screen of that resolution?
I could understand if the Linux boxes were running a powerful simulation or something, but surely we can get devices of that resolution running on a single PC with some pretty intense hardware.
Or, is GeForce 4 et al really on the cutting edge? What do the people with millions of dollars use? Do they have to start using multiple machines like IBM? Sounds unlikely to me. What about the military? Surely someone is one step ahead of the latest consumer technology?
mogorific carpentry experiments
Take one of thease
Add a little, traveling a sales man
and a few neurons
thank God the internet isn't a human right.
Why, do you have performance benchmarks comparing the experimental, non production DeepView with the commercially available InfinitePerformance visualization system?
Chromium is a project at Stanford. There was a paper published at this years SIGGRAPH which discusses how the T221 is driven by a cluster and the SGE, as well as other applications including a parallel volume renderer. (http://graphics.stanford.edu/papers/cr/) Chromium is an open source project and you can get it from http://chromium.sourceforge.net. Chromium is designed to enable people to harness the power of a "graphics cluster" and/or use multiple displays. You don't have to buy a T221 and an SGE to render Quake at high resolution, you can use multiple monitors/projectors instead. -Mike
I have seen this display driven by one Octane2.
a nn el.html
http://www.sgi.com/workstations/octane2/
Not one rack of 8 PCs and one half rack of some graphics engine.
It was drawing the full display at about 30fps.
It was as easy to use as any other workstation rather than the 'interesting' mix of 8 pcs in a rack and some other half rack of graphics stuff.
The hardest thing was to read the tiny little fonts on the screen (the display is 200dpi IIRC) - you need a magnifying glass (or of course you could increase the font size).
The Octane2 can do this because you can install two V12 graphics engines each with a dual channel adapter.
http://www.sgi.com/workstations/octane2/dual_ch
SGI software stiches the cards together transparently.
It is a beautiful display.
The amuzing/annoying thing about these sorts of announcements is that customers have been using SGI stuff to do this for the last 8 years or so.
If you were on the leading edge of this kind of work would you wait that long for some kludged together solution which might work if you have enough Duct tape to stick it together?
Or would you pay the extra cash for a solution that works and gives you a huge jump on your competition.
The sort of software layer that can be used to make these four channels (two channels from each graphics card) into one display is stuff like:
http://www.sgi.com/software/multipipe/sdk/
Oh yeah - and re your 'every table comment'. The Octane2 fits on one table - it doesn't surround the table like the IBM stuff.
Since Chromium allows applications to demonstrate pretty much linear scalability as nodes are added, I would guess that they got interactive frame rates. At least, that's what I've seen with Chromium displaying to one of those displays at Stanford.
SGI did the graphics cluster thing over two years ago and even released it as a product. Very, very similar to the IBM. They even ported several of their programming APIs and SDKs from InfiniteReality to the Linux Graphics Cluster. Not many were sold, however. Heck, Slashdot didn't even cover it. It was still pretty neat to see multiple spanned monitors and even composited high res projectors driven by a half rack of Linux PCs. Many of the demos were actually ported from SGI's big iron Onyx machines and worked just as well on the cluster. The basic setup was a stack of rackmount Linux boxes using nVidia AGP cards and custom PCI cards daisy chained together to provide sync for glSwapBuffers among other things. Also availble were gigE and Myrinet for networking the machines with something better than 100BT. A compositor (similar to what's used in InfinitePerformance) was also available.
m l
.edu and .org have also rolled their own graphic clusters... though I don't know who supplied the compositors.
More information (white paper and data sheet) can be found on SGI's legacy systems page:
http://www.sgi.com/products/legacy/vis_systems.ht
I belive a few
Way to go IBM! Now let's see you put it to a good use... say...
.edu would build this excellent... um, tool!
CAVE Quake!!
: )
Now if only my
NCSA has been making good progress on developing this. The million bucks of SGI hardware has of course been replaced by a rack of Linux PC's. Instructions are actually on line. It's not trivial to build, though. And using LCD projecters does have real downsides. If you buy a dozen identical projectors, they won't have the same brightness and color saturation. Actually, they aren't even consistent from edge to edge. So your display isn't perfect, and you can definitly see the tile edges. Not to mention the fun of building a usable support structure which lets you get all the alignments right. If you don't need it to take up an entire wall, I think the IBM T221 display is a cheaper way to get super high-res output. But of course your high-res Quake won't be lifesize either. :-)
Most of IBMs big machines seem to run on PPC chips of one form or another, and 866 is a common speed for G3 CPUs.
Deep Blue ran on the old PPC604e's!
I would think they are using IBM processors.
-- if it was so, it might be; and if it were so, it would be; but as it isn't, it ain't. That's logic - Lewis Carrol
IA-64 Itanium perhaps? However, looking at Intel's specs I don't see 866 listed as an available clock speed. They list the original Itanium at 733 and 800 MHz, and Itanium 2 at 900 MHz and 1 GHz.
I would have thought that IBM would plug their own hardware whenever possible -- the T221 display is certainly phenomenal, and they provide a link so you could buy one of those... It has me wondering how I can come up with $8400 to get one. (Heck, when those things came out, they were $20,000! Ah, progress.)
So, this leaves us to wonder... no mention of processors, "low" clock speed -- compared to what we're used to seeing -- something new from AMD? IBM pissed off at Intel? Some new massively parallel top secret silicon from IBM?
Watch... it'll turn out to be Pentium IIIs -- they call 'em "workstations," so they might have recovered them from some other project. (Or all the engineers got new workstations and wondered what cool project they could do with their old ones... Q3A at 3840x2400? What the heck!) What's the limiting factor in this case, processor power or network bandwidth?
"...America's great minds of today, teaching America's great minds of tomorrow. Poor bastards." -- A Beautiful Min
I would venture to guess because the Power4 processor is not optimized, nor designed to be used in a workstation environment. I would also suspect that it is not designed to be used as a graphics processor.
Power chips are not PowerPC chips. If you are refering to PowerPC Generation 4 processors, your message is a missleading question. Additionally, I don't believe that the PPCg4 is quite up to an x86 processor (off the shelf speeds) yet. Don't get me wrong, for the work I do, a G4 Mac would probably be sufficient, but if you are talking raw graphics processing, that's a different market.
-Rusty
You never know...
Apple used to sell an 867MHz G4 from July 01 - Jan 02.
That's a Mototola part of course, but I'm sure IBM has a similar PPC CPU.
-- if it was so, it might be; and if it were so, it would be; but as it isn't, it ain't. That's logic - Lewis Carrol
Wow! For just 2.5 times improvement I need an entire rack of Linux boxes.
-- SIGFPE
And I ought to add that my card does 4096x1536.
-- SIGFPE
With OpenGL spanning both screens.
-- SIGFPE