|Use the stack less :) email@example.com (2004-04-15)|
|Re: Use the stack less :) firstname.lastname@example.org (Diego Novillo) (2004-04-15)|
|Re: Use the stack less :) email@example.com (SM Ryan) (2004-04-15)|
|Re: Use the stack less :) TommyAtNumba-Tu.Comfirstname.lastname@example.org (Tommy Thorn) (2004-04-21)|
|Re: Use the stack less :) dSpam@arcor.de (Dietmar Schindler) (2004-04-21)|
|From:||Tommy Thorn <TommyAtNumba-Tu.Comemail@example.com>|
|Date:||21 Apr 2004 00:45:37 -0400|
|Posted-Date:||21 Apr 2004 00:45:37 EDT|
SM Ryan wrote:
> It's a well known problem that stacks are more difficult to pipeline
> because the stack top becomes a resource contention. That's part of
> why zero address machines like the old Burroughs or HP 3000 have given
> way to two or three address machines.
Sorry, but that makes no sense to me at all. A modern out-of-order
execution would immediately rename the "stack registers" and thus there
would be no contention in the execution pipeline.
The death of stack machines probably have more to do with being hard to
write optimization for and very hard to paralize. When code density is
truly critical (practically never) stack machines can be amazing dense.
Check out Bernd Paysan b16 Forth processor for an impressive example:
Return to the
Search the comp.compilers archives again.