Assembler backend (was: GNU Sather Status)

[Norbert Nemec]

On Thu, Mar 22, 2001 at 09:25:06AM +1200, Keith Hopper wrote:
>      Further work is going on here and elsewhere developing new libraries. 
> We would like to get started on a new compiler to get around some of the
> horrendous inefficiencies caused primarily by using the C progrmming
> language as an intermediate step to binary and having no code evaluator to
> save many thousands of run-time instructions being executed.

Just thinking about it: Are you sure, that a assembler back-end would be the 
way to go? I agree that the current C-code is inefficient, but it might gain us 
much more to try to output better C-code that can be optimized by the 
C-compiler.

At the moment, it should be our highest goal, to create a Sather implementation 
that is efficient and generally usable. Highly specialized implementations (And 
a assembler backend is to be considered "highly specialized") may then well 
follow if there are people who are interested in that specific problem. Sure, 
nobody can stop you from playing around with a assembler backend and testing 
it's speed, but that work will not really contribute much to Sather itself.

So, the way I propose would be to create a modular compiler with a C-backend 
which can then be optionally replaced by a assember backend.

Ciao,
Nobbi

-- 
-- ______________________________________________________
-- JESUS CHRIST IS LORD!
--          To Him, even that machine here has to obey...
--
-- _________________________________Norbert "Nobbi" Nemec
-- Hindenburgstr. 44  ...  D-91054 Erlangen  ...  Germany
-- eMail: <address@hidden>   Tel: +49-(0)-9131-204180