How crazy a machine can GCC usefully target?

Richard Henderson rth@redhat.com
Wed Feb 20 14:15:00 GMT 2002


On Wed, Feb 20, 2002 at 12:08:19PM -0500, Alan Lehotsky wrote:
> - completely exposed pipeline latencies.

This would have to be delt with in the assembler or 
machine dependent reorg.

> - No memory.

Ugly, but doable with a "register indirect" addressing mode.

> - No stack.

This is the killer.  There's no chance of making this work.


r~



More information about the Gcc mailing list