Robert Reimiller wrote: > On Sat, 20 Aug 2005 08:00:08 -0700, peter green <[EMAIL PROTECTED]> wrote: > >> dunno but from what i can gather softfloat code is generally faster >> than os >> level emulation because of the costs of trapping cpu exceptions in >> the os. > > > That's what I have heard too. I suppose if there is already an > emulation software > library that would be preferable. > > I think though I'm screwed as far as the endian issue is concerned. I > just checked > out the source using svn and found the line: " endian > : endian_little;" > for the Arm-Linux target definition. I kind of doubt that changing that > to endian_big and > recompiling will work
Give it a try. If it doesn't work, it should be little effort to fix it, but you've to recompile everything rtl etc. to get it work. About the softfloat stuff: I'll look into it the next days, it seems we need it for wince too. > > It turns out that Arcom's PXA255 board runs in little endian mode, > although it only > has USB 1.1 which would impact the disk performance (we have to use a > USB to IDE adaptor). > The Arcom linux development software actually comes with either > arm-linux-xxx (such as arm-linux-gcc) > or armbe-linux-xxx (for big endian) programs. I could use GCC, but I > don't have any > real interest in downgrading 35000 lines of Pascal into C. > > Actually I wish I could avoide the whole ARM/Linux thing altogether and > find a low-power > 586 class board with an OS that emulates Win32 so I could just make a > console application in > Delphi and run on it. Oh well. > > _______________________________________________ > fpc-devel maillist - [EMAIL PROTECTED] > http://lists.freepascal.org/mailman/listinfo/fpc-devel _______________________________________________ fpc-devel maillist - [EMAIL PROTECTED] http://lists.freepascal.org/mailman/listinfo/fpc-devel
