Re: faster lilypond rendering

[Richard Schoeller]

Josiah,

You got to this response before I could.  If the problem is swap
thrashing, then having 8x the memory to handle a score that is 3x the
pages should be plenty.  I had already disabled point-and-click.  It
helped.  The idea of using the PS output for proofing and going to PDF
only when I have it right makes some sense.  I'll have to fiddle with my
makefiles to distinguish the targets.  But it should help.  Thanks for
the suggestion.

It still would be nice if ps2pdf would actually manage its memory a bit
more subtly.  Converting a page, writing to the file and then returning
that space to the heap before starting the next page would be nice.

Dick

On Wed, 2006-02-15 at 12:35 -0800, D Josiah Boothby wrote:
> > I produce a 70 pages conductor's score in less than half an hour (and 
> > this is a *very* conservative estimate) on a two years old PC. True, it 
> > has 2GB of memory, but other than that, it is to be considered a fairly 
> > old beast by today's standards.
> 
> ...
> 
> >> And it takes forever on my 700Mhz PII!  So, cutting out processing of 
> >> sections of music is of little use.  I'm really looking for faster 
> >> rendering in general... The final phase of rendering a 28 page 
> >> conductor's score takes hours.  This is much more than twice as long as 
> >> other scores that are approximately half the length.
> 
> ...
> 
> >> FWIW, this could be caused by or at least exacerbated by poor memory 
> >> usage in the conversion phase.  If the PS to PDF conversion loads the 
> >> whole document into memory at once, this could all be the result of 
> >> swap thrashing.
> 
> A two-year-old PC with 2G ram is not going to be at all equivalent to a 
> 700 Mhz PII, no matter how much ram is being used. As suggested, the small 
> amount of memory is probably the culprit. It may be helpful to look 
> through the manual or the list archives for how to disable the 
> point-and-click feature, which adds information to almost all of the 
> symbols. I seem to recall a thread or three discussing this, and if you 
> are able to disable the point-and-click, you will probably get faster 
> ps->pdf conversion (and faster processing in general).
> 
> Josiah
-- 
Dick Schoeller
mailto:address@hidden
http://schoeller.hsd1.ma.comcast.net/
781.449.5476