Re: Fwd: How will cairo influence GTK's performance?

[Billy Biggs <vektor dumbterm net>]

Clemens Eisserer (linuxhippy gmail com):

> The fact that pango could be responsible for the slowness I experience
> with mozilla and especially eclipse are proofen by the fact that
> low-level swt benchmarks show a compareable performance of primitive
> draing functions, only text is about 3-5x slower than on the win32
> platform. [...]

  I don't think mozilla uses pango for text rendering (yet?), nor does
it really use GTK+ widgets -- at least not in the way Eclipse does
(mozilla draws like them though using the GtkStyle API AFAIK).

  While text rendering (and, maybe more importantly, text measuring)
does seem measurably slower using pango than using Uniscribe on Win32, I
am not sure that the difference in performance is the most important
optimization for improving Eclipse.

  In the performance work I have done with Eclipse, while we have
improved thing and fixed many bugs and inefficiencies, I have not found
large user-visible performance differences between Windows and Linux on
my machines.  Maybe there is something specific about your hardware or
software setup that is making the problem worse?  Is it the way you are
comparing which is significant?

  -Billy