 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Slower in v10 than v9
Nick Winter <[email protected]> writes that -graph combine- is slower under
Stata 10 than it was under Stata 9.
Nick is right. To elaborate a bit, it is the same speed when combining graphs
that have been saved to disk, but it is slower when combining memory graphs.
Graphs now combine the same speed whether they are memory graphs or graphs on
disk.
In Stata 8 and 9 we played a computer-science trick to increase speed -- we
combined memory graphs by reference. The graphs were not recreated, the
original graph was simply referenced from the combined graph. The sub-graph
in the combined graph and the original memory graph were actually
one-and-the-same graph. They might look different because they were drawn
into a different space, but they were the same graph. When a graph on disk
was combined it had to be created.
In Stata 10 we have a graph editor. Had we continued to combine memory graphs
by reference and you edited the original graph, the edits would have appeared
in both the original AND the combined graph. The same would be true if you
edited the sub-graph in the combined graph, the edits would have appeared on
you original graph (the next time it was redrawn). That's kind of
interesting, but not what you wanted. So, in Stata 10 memory graphs must be
recreated when they are combined.
If Nick wants his speed back he can add the undocumented option -nocopies- to
his -graph combine- command. Memory graphs will then be combined by
reference, as it was in Stata 9. Nick should be careful about editing the
either the original or combined graph.
-- Vince
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/