https://bugs.freedesktop.org/show_bug.cgi?id=36689
Winfried Donkers <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #3 from Winfried Donkers <[email protected]> --- Tested with LibO version 4.0.0.3 on Windows XP with a Pentium Dual Core [email protected]: SUMPRODUCT took approx. one second, SUMIF less. Given the cpu and the size of the sheet, the results do not worry me. Tested with LibO version 3.6.5.3 on Windows XP with a Pentium Dual Core [email protected]: SUMPRODUCT took approx. one second, SUMIF less. Given the cpu and the size of the sheet, the results do not worry me. Tested with LibO version 3.6 on openSUSE 12.2 with a AMD FX6300 6 core @ 3GHz: both functions were too fast to measure with stopwatch. I finally tested with LibO version 4.1 alpha+ (code update this weekend), build with full debug information and producing traces on modified code in calc, on same openSUSE12.2 machine: SUMPRODUCT took approx. two seconds, SUMIF less than a second. Given the build type, these results do not worry me. I looked at the code of the two functions: -SUMProduct is a matrix fucntion and calculates the summation of the product of (in your case) 2 matrices of identical size (13819x1), and one matrix is filtered before it is filled. -SUMIF calculates the summation of (in your case) 4510 products of 2 cell values, after 13819 rows have been filtered. These formulas are coded to handle much more complex contents (nxm matrices, with references, formulas, different data types, etc. ) and are fundamentally different, I am not surprised that SUMPRODUCT is slower, nor am I worried by the time the calculation takes. @Kevin: Is this still a bug for you? -- You are receiving this mail because: You are the assignee for the bug.
_______________________________________________ Libreoffice-bugs mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs
