Antialiased Fonts Acceleration

sshachar at sshachar at
Sun Sep 23 05:14:03 PDT 2007

Hi all,

In my system, the CPU is extremely overworked to handle non graphics related
tasks. I can however use a coprocessor for this purpose. It is a requirement to
run a modern X desktop with composite and especially antialiased fonts.

The catch is that since my coprocessor is, unfortunatly, a 32bit machine without
means for byte granularity manipulation I am finding it prohibitively difficult
to accelerate what is AFAIK the only API for antialiased font rendering and
that is EXA's: (e.g.) {solid 8bpp, composite add A8+A8=>A8, composite over
XRGB32+XRGB32(1x1 repeat normal, mask with A8)=>XRGB32, ..}

My question to you is this:
A. Is there some way of forcing the use of ARGB32 masks for fonts instead of A8
ones? Ideally, I need all operations to be on 32bpp formats
B. Is there some other API in XOrg with which I could transfer control of pixel
formats to use to the driver/HW?
C. Any other wise advice? (sigh)


More information about the xorg mailing list