Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!jato!hanauma.jpl.nasa.gov!hyc
From: hyc@hanauma.jpl.nasa.gov (Howard Chu)
Newsgroups: comp.windows.x
Subject: Re: Does anyone have gcc for the i860?
Message-ID: <1991May15.001354.8483@jato.jpl.nasa.gov>
Date: 15 May 91 00:13:54 GMT
References: <367@nic.cerf.net>
Sender: news@jato.jpl.nasa.gov
Organization: SAR Systems Development and Processing, JPL
Lines: 44
Nntp-Posting-Host: hanauma.jpl.nasa.gov

In article <367@nic.cerf.net> jonesm@nic.cerf.net (Mathew Jones) writes:
>I am a devotee of hopeless causes, so I am trying to compile the
>MIT distribution of X 11R4 in order to end up with an X server on
>an Alliant fx2800. My first attempts have met with failure because
>the Alliant C preprocessor, scc0, doesn't paste tokens together in
>the fashion expected by the X developers, and I end up with thousands
>of compiler errors. The X documentation that comes with the release
>says the gcc is the compiler of choice, so my question is...

I've recently finished building an X distribution on our FX/8s, (clients
only, got nothing to run a server for... This is slightly tangential
to your question but may be useful anyway.) On the 8[0]s, the pcc
compiler runs out of table space when compiling extensions/include/XInput.h.
It also doesn't allow large (8K+) local variables, it seems to have a
very limited user stack. I guess that's just FYI, I haven't used scc on
our 800 to see if it has the same limitations.

As another side-note, gcc may not really help you on the Alliant. On the
FX8s, the Alliant calling conventions aren't fully supported by gcc
(something incompatible in structure returns?) and what you wind up with
doesn't run. This problem probably wouldn't exist on the 800 since they
probably use a whole new calling convention, but ya never know...

The main thing I want to mention is that I spent some time rewriting the
Imake rules and template files to allow X to compile intelligently with
parallel makes. This cut my compile time on a wimpy old 4-processor FX/8
from about 14 hours to around 3 hours. (Perhaps I shouldn't quite say
"intelligently" up there, it drives the load average up into the 20s and
30s when it's chunking thru the lib and client directories... But the
Alliant Concentrix OS doesn't seem to get fazed too much by this, you really
don't notice any loss of interactive response...)

You will probably want to do the same thing... I think the default imake
rules need a lot of work anyway; fresh off the distribution, the thing
causes imake itself to be deleted and recompiled at every phase of the 5
step "make World" process, and this causes it to effectively repeat the
"make Makefiles" step 5 times as well. Talk about brain-damage.
(Personally, I think this is a plot by the CPU vendors in the XConsortium
to try to create demand for higher performance boxes... "Gee, why does it
take so long to build X?" "I dunno, I guess you need a newer/faster/more
expensive computer!"  }-)  }-)  }-)
-- 
  -- Howard Chu @ Jet Propulsion Laboratory, Pasadena, CA
	Disclaimer: How would I know, I just got here!