<html> <head> <base href="https://bugs.documentfoundation.org/"> </head> <body> <div> <a class="bz_bug_link bz_status_NEW " title="NEW - mdds accelerating random lookups" href="https://bugs.documentfoundation.org/show_bug.cgi?id=109097#c4">Comment # 4</a> on <a class="bz_bug_link bz_status_NEW " title="NEW - mdds accelerating random lookups" href="https://bugs.documentfoundation.org/show_bug.cgi?id=109097">bug 109097</a> from <a class="email" href="mailto:markus.mohrhard@googlemail.com" title="Markus Mohrhard <markus.mohrhard@googlemail.com>"> Markus Mohrhard</a> <pre>(In reply to Aron Budea from <a href="show_bug.cgi?id=109097#c3">comment #3</a>) > Created <a href="http://bugs.documentfoundation.org/attachment.cgi?id=134615" name="attach_134615" title="Callgrind trace">attachment 134615</a> <a href="http://bugs.documentfoundation.org/attachment.cgi?id=134615&action=edit" title="Callgrind trace">[details]</a> > Callgrind trace > > Here's a Callgrind output (well, the first of the 8 files). Unfortunately > the spreadsheet involved cannot be shared. Well, that one shows that the performance problem is in the matrix code where we use index based access to get each value out of two matrices and insert into a new one. We have existing code to do this with iterators which will improve the performance of that code significantly. At least from a quick look I don't see anything in this file that reflects the ScColumn::GetCellValue problem Michael mentioned. Based on a quick reading 75% of the import time is spend in 12 lcl_GetMinExtend calls with all the time spend in inefficient iteration through the matrix. Maybe you have a different document showing real problems but I doubt that this one corresponds to the problems mentioned by Michael. This one is actually a quite "easy" fix if you follow the existing ScMatrix performance hacks.</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the assignee for the bug.</li> </ul> </body> </html>