[rescue] Is anyone running netalk on solaris 9 sparc?

[Dave McGuire]

On Mar 14, 2004, at 11:34 PM, Devin L. Ganger wrote:
>>   One should bear in mind that, however, GCC's code generation for 
>> SPARC (and indeed most RISC) processors leaves a lot to be desired.  
>> If you have access to Sun's compilers, they will produce much faster, 
>> and in some cases smaller, binaries.
>
> I'd heard, and I don't have pointers to where now, that this was much 
> less of a factor with relatively recent versions of GCC.  Is that in 
> fact the case?

   I've heard this too, but I heard that it was "somewhat" less of a 
factor...not "much".  I haven't done any benchmarking against the 
latest release of GCC.  Sun's compilers are very, very good, so there's 
little motivation to keep trying GCC to see if they've figured out how 
to properly generate SPARC code.

   It seems that it's great on many CISC platforms, and indeed I 
remember truly dramatic increases in the performance of X11R4 on a VAX 
running Ultrix when moving from DEC's compiler to GCC, and a similar 
speedup when moving from Sun's compiler to GCC on a 68K...but (of 
course) those architectures are very different from RISC designs.

         -Dave

--
Dave McGuire          "PC users only know two 'solutions'...
Cape Coral, FL          reboot and upgrade."    -Jonathan Patschke