On Wed, Apr 26, 2017 at 11:00 PM, Kim Gräsman <kim.gras...@gmail.com> wrote: > > I've used a spreadsheet internally to forecast how much memory SQLite > will need in a given configuration. > > I've cleaned/generalised it a little and posted here: > https://docs.google.com/spreadsheets/d/1cCawZE9AdF4jlmgAMjbz_pGZI0o9WkWddPyV2YglFCY/edit#gid=0 > > Since we now get out-of-memory with a fixed heap, I'm guessing there's > some kind of think-o in the calculations above.
Let me confirm that. I'd completely neglected to perform the actual Robson calculation... So the totals I'd calculated were just a detailed forecast for the M part of Robson's proof. I still need to scale it by n to get a proper heap size, N: https://www.sqlite.org/malloc.html#_mathematical_guarantees_against_memory_allocation_failures Sorry about the noise. This is actually simple to add to the spreadsheet, so I'll extend it to cover Robson end-to-end. - Kim _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users