Re: lcc intel backend?

Gavin Thomas Nicol (nick@nsis.cl.nec.co.jp) wrote:
I should note that from the lcc papers I've read, it seems like it is much
faster, and much smaller than gcc, but code generation is not as good. ...


Not necessarily -- perhaps the opposite.


dobrien@seas.gwu.edu (David O'Brien) said:
> I am assuming the papers you are refering to are the 3 papers
> included with the anonymous ftp distribution. If I remember
> correctly they compaired lcc to gcc v1.x. Does anyone have any
> idea how the current version of lcc compaires with Gcc version 2.4.5?


Well, lcc *does not have a code generator*, for the x86 or any
architecture; it is purely a front end. Several back ends can be bolted
onto the front end. TWIG is the oldest. Several people are working or have
worked on other backend; frontends that spring to mind are BURG, IBURG,
and CHOP. All are described in the catalog of compilers recently posted.


So, you cannot compare lcc to GCC, as far as code generation is concerned;
one should code generated by GCC to that produced by one of the backends
that can be attached to lcc. As far as frontends are concerned, the lcc
frontend is way smaller/faster/more maintanable than the GCC frontend.


Some time ago Dick Dunn and another guy asked me about all this saying
that if they found the time he might do an x86 backend for lcc using
perhaps one of the above tools. I never heard from them again -- if they
are reading, perhaps they might want to comment.


I thinks that doing an x86 backend for lcc using iburg or chop might be a
fun MSc level work -- any takers? (I cannot, I am trying to develop the
next Great Western Operating System :->).
--

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