View Full Version : Viewport speed and Video memory
greyface 03-06-2006, 01:37 AM Hey, I'm wondering how important video memory is to viewport performance (OpenGL) in a 3D app. Any insight? Specifically, 128MB vs 256MB.
Alex-
|
|
imashination
03-06-2006, 01:24 PM
Hey, I'm wondering how important video memory is to viewport performance (OpenGL) in a 3D app. Any insight? Specifically, 128MB vs 256MB.
Alex-
The memory will mostly make no difference to speed. Only once you load more textures and geometry than will fit on the card will it make a difference. Most 3d apps will only store the textures on the cards memory, doing most geometry transformations in main memory. The bottom line is that the gpu will make a much larger difference than the memory will.
greyface
03-08-2006, 11:45 AM
Thanks for the info. But are you saying that geometry can be stored in vram? Why is it that high end cards have enormous amounts of video memory?
Alex-
Textures......
But if your game does not fill up the texture buffer, getting a card based on how much RAM it has in comparison to your current, is a bit silly, since you wont really see an improvement due to the extra RAM that can be devoted to textures. However, if the RAM is faster then you are likely to see a difference...
If you plan on running games at thier absolute maximum settings, RAM will make a difference, as calculating anti-aliasing in real time often requires larger amounts of RAM than usual. On top of the higher resolution textures used in higher display settings. Then there's the shadow and lighting info, etc, etc..
greyface
03-08-2006, 01:20 PM
This is for content creation though, specifically modeling, not games. So I question the use of having big amounts of memory like the quadro fx 4500 has, for example if it's just for textures. When modeling or animating, you don't often need to see the textures..
Indeed, which leaves 128 or 256 MB as a decent number. Most cards will have at least 128. Most modern high-midrange cards will have 256 MB, while the midrange and lower usually have 128. If low range cards have 256, it is often slower than the 128 solutions, and you'd probably be better off grabbing one of the 128s insted.
But honestly I'd be more concerned about the speed of the RAM and what GPU you are buying, RAM wont give a big impact unless you need HUGE models and big textures all in real time as you animate. For example, having all of the high quality settings enabled in Lightwave via the recently added GLSL functionality, will probably make use of some of that RAM, but if you dont need to see textures, fog/lense effects in the UI, and would rather render them out to see how they behave, I dont see big RAM helping much (512MB and up).
However dont hold the RAM size of the Geforce 7800GTX 512 against it, as this card not only has more ram but faster memory and GPU speeds than the standard 256 version. So speed is coming from the increased clock speeds.
greyface
03-10-2006, 01:47 AM
Alright, gotcha - thanks for the help, I understand better now.
CGTalk Moderation
03-10-2006, 01:47 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
vBulletin v3.0.5, Copyright ©2000-2013, Jelsoft Enterprises Ltd.