[XeTeX] performance on Windows machines
jonathan_kew at sil.org
Mon Sep 11 20:32:42 CEST 2006
On 11 Sep 2006, at 5:47 pm, William Adams wrote:
> I've killed a second HD in my Windows pen slate, so am contemplating
> getting a replacement machine --- does anyone have any perspective on
> how much faster a Celeron-500 will be over a Pentium/MMX-233 for
> running XeTeX? Any hope of it being significantly more than twice as
> fast as the MHz comparison would indicate?
[somewhat off-topic response]
I have no idea about that comparison, but I can tell you it really
flies on a new Mac Pro!
To test performance with legacy TeX stuff (not OpenType, etc), I
sometimes run texbook.tex (with a suitable prefix file to make it
actually typeset). The Mac Pro takes about 1.1-1.2 seconds (real
time) to generate the 494-page PDF using xetex -output-
driver="xdvipdfmx -q -E". Wow! Actually, that's faster than pdftex
can do it. :)
Factors that may help performance include disk speed as well as CPU
speed. That's probably more important than having lots of RAM, as
xetex isn't all that memory-hungry (by today's standards). Having
dual CPUs helps, too -- that's why xetex can beat pdftex on the Mac Pro.
Note that typesetting with OpenType fonts is a lot slower (by a
factor of several times, usually) than using TFMs; while there's
certainly scope for optimization, this is partly due to the much more
complex glyph layout process, so it will probably always be true to a
large extent. That's the price you pay for the flexibility and power
you get with the richer font model.
More information about the XeTeX