Keep 3DLuVr online!
3DLuVr Logo
Sections
Articles
 From the Real World
 Digital Painting Series
 Featuring of...
 On the Bookshelf
Tutorials
 3ds max
 LightWave3D
 Softimage XSI
 Rhinoceros 3D
 Video Tutorials
FunZone menu
 I always wanted to be
 Talk to an employer
 Why Ask "Why"
TechZone menu
 Hardware Reviews
 Software Reviews
 Benchmarking
 Q&A, Tips & Tricks
UserZone menu
 The Artist Sites
 15 Min of Fame
 Request an Account
 Current Assignment
 Sponsors & Prizes
 Make a Submission
 Voting Booth
 Competition Rules
About menu
 Mission Statement
 Policies
 Advertising
 Comments
 Poll Archive
 Links
 How to IRC
 Donations
Login
Log in to be able to post comments to the news items, forum posts, and other facilities.
Username: 
 
Password: 
Not registered? Register!     Lost Password?
Poll
 Your New Year`s Resolution is...
Gain employment
Stop smoking/drinking/etc
Get back in shape
Find the meaning of life
Conquer the World
Absolutely nothing

    Poll Results
Comments
Want to leave us a comment about the site or in general? Click here to access the form.
TechZone Heading
Workstation Smackdown (Nvidia Quadro DCC review)
Added on: Mon Feb 04 2002
Page: 1 3 4 5 6 7 8 9 10 11 12 13 14 15 

The Heart of the Monster

The GPU, friend of the 3D artist. Capable of feats beyond mortal man, these mysterious device removes a huge amount of overhead from the system processor, processing it on the card itself. Commonly referred to as a “Transforming and Lighting” engine, this part of the Nvidia processor is responsible for its raw polygonal throughput abilities. It determines how many polygons you can throw at the viewport before you get the infamous chug effect.
(For those of you who don’t know the infamous chug effects, its similar to drinking a whole lot of beers at once.

First you get a little dizzy, your viewports start dropping frames, and then you usually fall.) The T&L (not the bar) engine free’s up the cpu for other tasks, like trying to get Max booleans to function properly. The Quadro’s GPU is theoretically capable of pushing a few million polygons a second, but these are all pure gross polygons (Not ugly, but unlit, unshaded, untransformed, unborn polys), and this number doesn’t really reflect actual performance. (Thus the whole point of this long winded, yet surprisingly humorous article).

All of this power fits into a tiny package which holds a scant few 57 million transistors. (20% more then a Pentium IV). What’s all these numbers and explanations mean to you, the end user? It means faster geometry manipulation in the viewport, smoother viewport animations, and greater poly management.
(Higher Polygon Ceiling) But all the power contained in this small .15 micron process would be useless without effective memory management. That’s where lightspeed comes in to effect. (Not a joke)

To Lightspeed and Beyond

One of the strong points of the Geforce 3 line (Quadro Included) its their incredible amounts of ram bandwidth. Using their patented “Lightspeed” technology, Nvidia is able to give their video cards poly crushing power, and the ability to quickly allocate critical ram resources to the crunching factory.
The Lightspeed memory technology works through the “Crossbar” controller, a set of four 64-bit controllers which help to reduce the latency of memory transactions.

( If you don’t know what latency is, try this simple experiment. Find a door in your house. Close it. Walk into the door. You might want to consider stopping before impact. Wait a second. Open the door. Now repeat this experiment, only this time, open the door as you get to it.
Reduced latency = less waiting. It is unknown at this time whether nvidia will patent this “less waiting” system at local Motor Vehicle Associations)

Here’s an example of how the controller works. On the left you’ll see a normal 256 bit controller and its corresponding four memory chips.
On the right (Quadro DCC) you’ll see the four 64-bit controllers. Red lines demonstrate communication between the entities.


As you can see in the images, the crossbar memory controller of the DCC allows for greater transmission of executed bandwidth, resulting is lower latency, higher bandwidth, and greater overall speed. Combined with the high speed DDR available to these controllers, the Nvidia line has an incredible amount of memory bandwidth to utilize, over seven gigabytes a second. This translates into higher resolution performance, in shaded and textured operations in the viewport, providing for less lag and more modeling power, under any circumstance.

But what is power without finesse. Letting loose a firehose on a pinhole accomplishes nothing. Unleash it on a crowd of [email protected] users and you have a riot.
Nvidia tackles this use of power with an innovative and powerful feature known as shaders. (Not sunglasses). Pixel and Vertex to be exact.



 
© 1997-2017 3DLuVrTM (Three Dee Lover)
Best viewed in 1024x768 or higher,
using any modern CSS compliant browser.