On Wednesday, 17. November 2010 08.23:22 Graeme Geldenhuys wrote: > I then did a simple test to see what could be the cause of problems in the > LCL app (and why it looks frozen). I created a stringgrid with exact > dimensions, a single column. I then used a simple test method that > generates log entries (8 million to be exact - did start small though). > Each time a log is received, I do a BeginUpdate, add a new row to the grid, > populate the first cell with a number, then call EndUpdate. Under fpGUI > this test took 5 seconds to complete, MSEgui took 10 seconds (just for fun, > I tied MSEgui too). LCL took.... well, I kill the app after 3 hours of 100% > CPU load and still nothing being displayed. Thank goodness I have a quad > core, so the 100% load on one core did not affect my day-to-day work. > You probably should give a little bit more information. ;-) The thread is here: http://msegui.org/fudforum/index.php?t=msg&th=20921&goto=136104&rid=0&S=c3696d6681449225b1fe14c93101f120#msg_136104
My results with MSEgui: " I get the following results with AMD Athlon XP3000+, 1GB RAM: (2GB in mail was wrong information) 1: Your testcase: 33.9s (this is the equivalent of your 10 seconds on my slower computer) [...] 2. Assigning a UnicodeString (no iconv calls): 16.9s [...] 3. No column index in loop: 16.3s [...] 4: Direct access to the column datalist: 3.54s [...] 5: Direct access to the data with allocating the memory once: 0.52s [...] With Unicode inttostr() I get 6.3s (trunk 4061): [...] " Martin -- _______________________________________________ Lazarus mailing list Lazarus@lists.lazarus.freepascal.org http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus