karl@giza.cis.ohio-state.edu (Karl Kleinpaste) (06/27/89)
I have received and minimally tested a port of GCC to the Pyramid which was done by Jonathan Stone <jonathan@comp.vuw.ac.nz>. He sent it to me and asked that I make it available for copying. You can pick it up via ftp from tut.cis.ohio-state.edu [128.146.8.60] in pub/gnu/gcc/gcc-pyr.tar.Z; or via UUCP from osu-cis as ~/gnu/gcc/gcc-pyr.tar.Z. The file is 43158 bytes long and contains: rw-r--r--482/10 1795 Jun 22 09:49 1989 gcc-pyr/README r--r--r--482/10 3871 May 28 22:59 1989 gcc-pyr/out-pyr.c r--r--r--482/10 34090 May 28 22:59 1989 gcc-pyr/pyr.md rw-r--r--482/10 704 May 28 22:56 1989 gcc-pyr/real.h.cdif rw-r--r--482/10 6648 May 28 22:55 1989 gcc-pyr/stmt.c.cdif r--r--r--482/10 47645 May 28 22:59 1989 gcc-pyr/tm-pyr.h rw-r--r--482/10 815 May 28 23:38 1989 gcc-pyr/va-pyr.h rw-r--r--482/10 536 May 28 22:56 1989 gcc-pyr/varargs.h.cdif r--r--r--482/10 1550 May 28 22:59 1989 gcc-pyr/xm-pyr.h In order to build GCC successfully, you'll need to be running a Pyramid configured with rather a lot of swap space. Both of my OSx4.4 Pyrs have only a single `b' swap partition; when trying to build stage2, with the machine otherwise essentially idle and about 20Mbytes of VM space available (as reported by pstat -s), compilation of several files still failed - GCC is an alarmingly memory-hoggish compiler. I had to revert to one of my OSx4.0 systems which has about 115Mbytes of swap configured. This was on a 98x, and the resource statistics reported by csh at the completion of the 2nd and 3rd stage build procedures were: 2934.2u 965.6s 3:24:33 31% 109+669k 6561+7700io 5858pf+1w 2417.9u 902.1s 3:46:23 24% 107+136k 6816+7645io 8678pf+0w I could only successfully get through the 1st stage on our 9825 due to the swap limitation, but even so, its stats were: 335.1u 62.7s 10:02 66% 46+305k 1556+1698io 575pf+0w (Realize also that the 9825 is relatively "quiet" compared to our overworked 98x...for now.) Plan on a very long lunch when you start building. --Karl