Hi Przemek,
> > 2009-03-30 13:10 UTC+0200 Viktor Szakats (harbour.01 syenar hu) > > * INSTALL > > * Split Windows compiler list to 'recommended' and 'supported' > > sublists. Recommended are MinGW and MSVC. > > Some comments on the rest: > > - pocc: lack of 3rd party support, mediocre performance. > > - owatcom: lack of 3rd party support, bad performance, > > proprietary object/dll format. > > You killed the performance by setting -bm flag as default. > When -bm is used OpenWatcom uses some different memory allocator > which is extremely slow. The next switch which probably reduce the I didn't want to kill it, just wanted to tweak it for MT.. we're using such MT switch for most compilers to force MT friendly runtine/code generation/ whatever. Isn't this needed for owatcom? --- -bm build target is a multi-thread environment --- --- bm (Netware, OS/2, Windows NT only) This option causes the compiler to imbed the appropriate multi-thread library name in the object file. The macro __SW_BM is predefined if bm is selected. --- > performance is -5s but I haven't made any real tests here and I only > guess that stack calling convention is slower then register one. It's definitely slower, but needed for compatible .dll generation as far as I understood Andi's mail. Didn't spend time to do side-by-side comparison though. > Anyhow here the difference will be probably minor. > Without this switches OpenWatcom gives the fastest code in few places. > F.e. even ICC is not close to OpenWatcom results in T029 and T030. > The overall performance is also quite good. In my tests it was noticable > better then BCC or POCC. Great, so if you say -bm isn't needed for MT support I can remove it. Shall I go for it? Unfortunately owatcom has the same problem as bcc/pocc: lack of support, so some libs had to be disabled due to compile errors (libhpdf, hbqt), but I otherwise I have no problem keeping it in first line. Brgds, Viktor
_______________________________________________ Harbour mailing list Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
