stefancoe
stefancoe
@jpn-- On the call I reported that I was still seeing the issue that @dhensle posted above. I wondered if it had to do with using the same output directory...
> > I think the functionality should be updated to delete old timing files before writing new ones. > > I imagine users potentially want to keep older expression timing...
@jpn-- I can also do a runtime test. My tests so far have been with our small set of inputs, which runs in about 12 minutes, single process. I did...
Moving to Pandas 2.0 could (should?) provide performance gains: https://datapythonista.me/blog/pandas-20-and-the-arrow-revolution-part-i
It looks like multiprocessing was added: https://github.com/ActivitySim/populationsim/pull/130
It seems like the CSV file cannot have a header (field names)? For example, it would be nice to be able to specify the columns for the source, target, and...
> > > Also, for reference we benchmarked a while ago that the time to load a graph with all of the USA roads (~23.9M nodes, ~58.3 M edges) took...
One reason is for aggregating model results up to non-nested geographies. For example, it's probably ok to aggregate MAZ results to city or other political boundaries by using the MAZ...
We will still use these tables though. I am mostly asking as a way to get the model to run during development.
Thanks @i-am-sijia! I tested running with tracing off and it did not crash so we are able to compile sharrow. I also tried testing with sharrow and it does not...