On Sun, 2013-03-03 at 20:35 +0100, Håvard Berland wrote: > Bård Skaflestad: > > I personally think a fixed-size buffer (e.g., one megabyte) which is filled > > through fread() and then > > separately tokenised is a much better choice for the lowest > > level of a deck parser. > > I hope you are not suggesting that one megabyte ought to be enough for > all reservoirs...
Not at all. If I were to make such a suggestion I would have chosen a more appropriate size. 640 MB springs to mind. My one MB suggestion was an order of magnitude intended to balance memory consumption with input performance. I'll happily go for five or ten, but I don't think we need to go for buffers that are much larger than that. On the other hand, measurements trumps assertions every time. Naturally, there will be looping involved (until 'END' or EOF), and possibly one buffer for each (open) file in the INCLUDE structure. Bård _______________________________________________ Opm mailing list [email protected] http://www.opm-project.org/mailman/listinfo/opm
