[cairo] [PATCH] Rely less on fast double-precision FPUs
cairouser at yahoo.com
Tue Jun 1 17:01:43 PDT 2010
Probably because of range? Cairo has enough trouble with large values
that do not fit into the "fixed point" representation. Limiting them
further to 32 bit float would make life even harder. Some of us draw
using really really large numbers (with surface offset appropriately
shifted :) ).
Bill Spitzak wrote:
> I'm sure this has already been discussed for cairo, but is there a
> reason double is used instead of 32-bit float? Many other graphic apis
> I have seen use float coordinates and matrices for precisely this reason.
> Jonathan Morton wrote:
>> ARM CPUs have a big weakness in double-precision floating point. In
>> particular, DP FP compares are ridiculously slow, mostly because the
>> result of the compare has to be transferred from the FPU to the CPU
>> using a serialising instruction.
> cairo mailing list
> cairo at cairographics.org
More information about the cairo