[OS X TeX] TeX Performance (iBook)

[Alexandre Enkerli]

Interesting thread. Couple of things to note.
Specs:
iBook (Dual USB) 500MHz, 384MB RAM, Mac OS X 10.3.7, GW's teTeX, 
standard drive
Yes, a slow machine from 2001. Usually, this machine is perfectly fine 
for day-to-day TeXing, especially as, for most documents, the actual 
typesetting is only done once the document is mostly ready.
An advantage of a slow machine for testing performance is that 
differences in performance are more noticeable.

TeXing the source2e document takes some time to generate fonts on the 
first pass so these results aren't that reliable:
real    1m8.796s
user    0m49.160s
sys     0m6.940s

It seems running on battery has no effect if the Energy Saver options 
are set to maximal processor speed.
Plugged in:
real    0m45.729s
user    0m40.340s
sys     0m2.230s
Battery:
real    0m45.709s
user    0m40.130s
sys     0m2.220s

However, impact of running with the reduced processor speed option is 
unsurprisingly noticeable:
real    0m53.869s
user    0m48.960s
sys     0m2.610s

Will try to do a test booting from a much faster external LaCie drive 
(80GB, d2 enclosure). In the past, it has made a very significant 
difference in processing speed.

In fact, one advice for those wanting to use the Mac Mini as a 
TeX-machine is to use an external drive. True, it makes the set up 
bigger and might be a fairly big investment. But it also has several 
advantages including being able to swap the drive between different 
machines, easier backups, and having more drive space. Plus it requires 
no tinkering to install... Granted, external drives may not be silent 
but it doesn't seem Steffen's query was for a silent machine.

Did anyone do tests using their machine in pure CLI (say in "Single 
User Mode")? It seems that, at least for RAM, the Finder and 
Aqua-related threads are somewhat hungry, which might have an influence 
on processing speed.

According to the Activity Monitor, pdfetex seems to be using between 
80% and 90% of the processor with some speed given to AM itself but 
that doesn't affect the "user" time. With AM running:
real    0m49.648s
user    0m39.600s
sys     0m2.570s

Unsurprisingly, disk access is mainly output and occurs in short tall 
bursts (in AM). Could a RAM disk be used in this case? If so, how 
(i.e., what needs to be put there)?

Then, there are procedures to make things faster in general, for 
instance only typesetting changed sections (either with \includeonly or 
more specific techniques).

All told, 40s to typeset 437 pages isn't *so* bad, especially as 
typesetting the whole document should be a fairly infrequent task. 
Granted, that document doesn't have very demanding typesetting 
features. Anyone knows how much of an impact XeTeX-specific content, 
for instance, would have on processing time?

Alex

Alex Enkerli, Teaching Fellow, Visiting Lecturer
Department of Sociology and Anthropology, Indiana University South 
Bend, DW 2269
1700 Mishawaka Ave., South Bend, IN 46634-7111
Office: (574)520-4102
Fax: (574)520-5031 (to: Enkerli, Anthropology)

--------------------- Info ---------------------
Mac-TeX Website: http://www.esm.psu.edu/mac-tex/
           & FAQ: http://latex.yauh.de/faq/
TeX FAQ: http://www.tex.ac.uk/faq
List Post: <mailto:MacOSX-TeX at email.esm.psu.edu>