> OK, it's true that Linux has some brilliant programmers available that
> could work full-tilt on putting together a Linux distribution in record
> time. But the SPARC port is not the best basis of comparison with the
> 1) The SPARC architecture has been around for quite some time, so many
> people have had a chance to program in SPARC assembly language.
At least one third of the UltraPenguin hackers did not have any SPARC
experience one year before. He learned SPARC assembly by looking at
the output of cc -S. This same guy is also responsible for Linux
having the fastest memcpy operations on the SPARC.
> 2) The machines are quite common by now, specially in academic settings.
> It's quite easy to find a SPARC machine where one can get a user
It would be much easier to get access to a merced machine.
> 3) GCC has been ported to the SPARC for quite some time.
The UltraPenguin team had to do some extensive hacks gcc/binutils to
get the thing up and running on the SPARC-v9 which is the 64-bit
extension of the SPARC architecture.
Porting the GNU gcc/binutils to a new platform is not a difficult
task. If 3 hackers could get the Ultra-64 port up and running before
Sun could port Solaris to their SPARC-64 chip, imagine what thousands
of Linux hackers can do with the intel 64 bit chip.