SLOW v.digit2, non-working v.digit
mccauley at ecn.purdue.edu
Fri Aug 20 16:08:11 EDT 1993
Ronald Thomas (ront at niwot.CFNR.ColoState.EDU) writes on 20 Aug 93:
>Will there be any help forth coming on a solution to these problems, or do
>we "pioneering" GRASS4.1/Solaris users have to wait for CERL to switch to
>Solaris 2.x before these things are looked into? Are these GRASS4.1
>problems, or are they Solaris 2.x problems?
best I can figure, from watching vmstat and from observations of other
software, I estimate that the problem is 95% Solaris 2 and 5% v.digit.
The new v.digit is an excellent idea but it came at a really bad time
w.r.t. Solaris 2.
For any enterprising soul who wants to check this out, I would suggest
profiling the code (v.digit2/altek) and find out where the bottlenecks
are occurring (see 'cc -p', 'monitor', 'prof'). This would give most
folks a temporary fix until we can get the new v.digit figured out.
I agree with Ronald - THE PROBLEM IS SERIOUS. We at Purdue are dead in
the water w.r.t. digitizing. I'm know that boothl at ecn.purdue.edu has
done specific timings of v.digit2 (he's our main digit-dog), but to
just give you a rough idea, I would estimate that it would take about
30 minutes to an hour (start to finish) to digitize a 3 inch tall copy
of the Corps' Castle in stream mode.
--Darrell (glad that my own research involves point data) McCauley
More information about the grass-dev