On Jun 10, 2005, at 1:55 PM, Rob Bearman wrote:
I've been building 1.4 on the Mac (against qt-mac-free-3.3.4) for
comparison
purposes but I haven't seen any glaring perf problems. Forgive me if
I'm
confusing the issue by not having followed this thread carefully, but
I've
run the latest build on a 512MB Mac Mini and a 768MB G4 Powerbook,
loaded up
the User's Guide, copied and pasted it a couple of times to make a
bigger
file, and find no discernable degradation in responsiveness when I
type in
random places. I was curious after having read that 1.4 Mac is
unusable. Am
I missing something?
Assuming not, I'd be curious to know what the memory configuration of
Bennett's machine is. Is it possible that the issue isn't one of code
perf
but of swapping?
Rob -
I have a 512MB iMac G4 (1.25GHz); it's surely not my hardware that's
the issue.
Perhaps it's the configuration. Here's what I've got (compiled with
gcc-3.3):
./configure --with-frontend=qt --without-x
--prefix=/Applications/LyX-140.app --enable-maintainer-mode
--with-included-gettext --enable-optimization=-Os
--disable-concept-checks --with-version-suffix
--with-qt-dir=/Users/bennett/lyx/qt-mac-free-3.3.4 --without-aiksaurus
(which means I'm also using qt-mac-free-3.3.4). Furthermore, I've got:
LDFLAGS="-framework Carbon -framework OpenGL -framework AGL -framework
QuickTime -lz"
Anything out of line here? Do you have something different?
Why do I say things are so slow as to be unusable? When I type into an
empty document, lyx starts off (according to top) using about 20% of
CPU. After I've finished typing 4 lines of text, CPU usage for lyx is
up to about 45%. After 10 lines of text, it's up to 75%. By 15 lines of
text, CPU usage is pegged at 100% overall, and text appears on screen
only after a delay -- a delay which increases the longer the document
is, at least up to a point. With a long document, when I type "The
quick brown fox jumps over the lazy dogs." (which takes me about 7 or 8
seconds), the delay goes up to 9 seconds *after* I finish typing before
all the text appears. The complexity of the document -- whether with or
without math, footnotes, bibliographical entries, cross-references,
etc. -- seems not to be a factor.
Bennett
