On Fri, 27 Nov 2009, David Arturo Macias Corona wrote: Hi,
> As a summary: > - OS/2-Watcom was failing with GPF in MT mode > - Tests lead us to devices input problem (keyboard, mouse, ...) on > threads except 1 Exactly. > - Maurilio made tests changing switchs and it work for him, removing -s > switch > - I made tests with switchs and removing -s switch does not GPF Can you repeat your test for current SVN code and check if removing still -s resolves the problem? > - But, some tests run slow, freeze, hangs, ... > - ... and here we are The main reason of slowness was -bm switch. It existed from beginning in all OpenWatcom MT builds and was caused by very slow OpenWatcom MT safe memory manager linked automatically with final application if code was compiled with -bm. As workaround we were using DLMALLOC in Windows. Unfortunately DLMALLOC was not working in OS2. I made some basic modifications to compiled it cleanly in OS2 but the OS2 port of DLMALLOC was never finished and if you tried to use it then GPF was generated. Recently with Maurilio help I finished this modifications and now DLMALLOC can be used in OS2 what should resolve the main reason of the slowness though removing -s causes some additional overhead which also reducse the speed, i.e. in Linux builds it change the speedtst result from: [ total application time: ]....................................23.94 to: [ total application time: ]....................................29.63 but it's only ~24% slower not ~300% and we can try to find some workarounds later. Now it's important to check if current SVN code works correctly in MT mdoe when -s is removed from build switches. Please make tests with MT version of speedtst with such build for: speetst.exe --threads and: speetst.exe --threads=2 --scale > So we faced two problems: > - GPF in MT related to devices input, solved? removing -s > - Slow, freeze, hangs, ..., un-solved yet See above. I need current code results. > A big note: > - Maurilio was using OS/2 SMP kernel > - I was using eComStation 1.2MR, OS/2 4.5, both tested in two machines I know but the problem seems to be inside OpenWatcom code and for me it looks like OpenWatcom bug which probably can be quite easy fixed but I need more information to be sure. Sorry that I'm asking you about it but I do not have OS/2 to make such tests myself. best regards, Przemek _______________________________________________ Harbour mailing list (attachment size limit: 40KB) Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
