← Back to team overview

kicad-developers team mailing list archive

Re: GAL rendering performance

 

Hi Jeff,

On 11/3/18 4:29 PM, Jeff Young wrote:
[snip]
> Are vertexes allocated per group?

Each group gets a contiguous memory area. This way a group is drawn from
cache (video memory) just by specifying offset and size of the chunk
storing the associated vertexes.

And is the whole text item a group?

Correct.

Would it be faster to reserve by character or by text item?
I think it might be just slightly faster to reserve all needed memory in
advance, rather than individually for each character. Getting a
contiguous memory chunk from CACHED_CONTAINER is rather cheap, unless
the container is so fragmented that the only forward is to squeeze all
stored data (defragment). IMHO the most interesting code parts to read
are CACHED_CONTAINER::{Allocate,reallocate} methods.

Cheers,
Orson

Attachment: signature.asc
Description: OpenPGP digital signature


References