Skip to main content
gamedev.net
🔒 Locked

hardware question...

Started by Zantac Nov 2, 2000 at 4:25 AM 2 replies 250+ views
Original Post
Zantac
Zantac
Hello! This isn''t as much a problem as it is a question... i''ve been playing around with the mipmap lod bias in d3d, and noticed that changing it had a SEVERE impact on the framerate. In my engine i got double the fps when i went from the smallest lod (cripsiest textures) to the largest lod (blurry textures). Why does it take the hardware (i''ve got a geforce2 mx by the way) longer to render larger textures? Shouldn''t it be all the same since all of the textures are in the card''s local memory? (i haven''t filled it up, so there''s no texture swapping from system memory). Is there anyway to get past the impact on the fillrate (because the game looks much better with crisp textures)? Thanks, Simon
Coder10000
Coder10000
It takes longer for a graphics card to render larger textures, because they need to access always the complete texture for filtering. Your Geforce 2 MX can''t access the RAM quickly enough to render everything as fast as with the small textures.

Coder10000
mhkrause
mhkrause
Larger textures take more memory, so it takes longer to swap them into vidmem.

Also, 3d cards have a texel cache. Smaller textures can fit entirely or mostly in the texel cache. Larger textures have more texel cache misses.

Zantac
Zantac
Ok, thanks for the explainations!

Cheers,
/Simon

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.

Sign in to reply to this topic.