Archived

This topic is now archived and is closed to further replies.

Dynamic buffers is the only right way to render big scenes

This topic is 5972 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I read something about vertex/index buffers in nvidia developer page. I understood that it''s better to use dynamic vertex and index buffer instead of many static buffers. Changing vertexbuffer is so time consuming task in D3D. Is that true? Actually it makes many things much easier. I can more flexible arrange my vertices and indices. It''s also important to have minimum vertex size possible so that copying vertices takes less time! Maybe FVF(POS/DIFFUSE/TEX)would be good? This is very interesting since I''ve been struggling with static vertex buffers for a long time. I always thought that locking dynamic buffer and copying vertices would take much more time than changing streamsource. What about doing all with dynamic buffers? I''d need just one vertexbuffer and one indexbuffer. I tried something: Copying 100 000 vertices took about 5ms. Not bad. And using dynamic buffer allows to fill buffer at the same time when hw is rendering prev buffer. That sounds promising Changing texture is much lighter than changing vertex buffer to render from??? How do they do it in commericial games?

Share this post


Link to post
Share on other sites
Whenever possible, you want to minimize the amount of data that has to be moved back and forth by the card. This usually means that you want to minimize locks.

5ms seems like a long time - you''re running at 200 fps just to move the data, and you haven''t even started rendering yet... 5ms seems like a large chunk of your per-frame time.

Can you provide the link to the nVidia article?

Share this post


Link to post
Share on other sites