Re: 32-bit vs. 64-bit x86 Speed

Jon Forrest wrote:
> Let's say you're a Linux user who never needs to run programs that
> don't fit in 32-bits. Would you run a 32-bit or a 64-bit version of
> Linux? You compiler people probably have intimate knowledge of the ISA
> issues here so I'm interested in what you have to say.


In a JVM you have a choice of how you represent the object references.
One technique is to used compressed references. Compressed references
prevent the problem that references make your objects bigger and waste
cache space, at some arithmetic/logic cost. There are some results for
this on BEA's web site:


http://edocs.bea.com/jrockit/releases/5026x/relnotes/relnotes.html#wp1078122


Also in relation to your question, there are issues with how well the
ISA maps to the microarchitecture. I believe early Intel EM64T
processors had worse performance in "long mode" than in regular 32bit
protected mode. I would imagine that now both modes map and perform
just as well. In recent optimization manuals Intel have deprecated the
use of the 32bit increment and decrement operations, as in long mode
these are used as prefix operations to address the 8 extra
registers. This would suggest long mode was becoming the norm, just as
protected mode slowly took over from the segment:offset way of viewing
memory.


In a 64bit Linux you always have the choice to compile and run
applications as 32bit. GCC provides the -m32 flag for this purpose.


Regards,


Ian Rogers

Return to the comp.compilers page.
Search the comp.compilers archives again.