[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] address@hidden: Re: -mcmodel=large

From: Camm Maguire
Subject: Re: [Gcl-devel] address@hidden: Re: -mcmodel=large
Date: Sun, 22 Sep 2013 09:37:26 -0400
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/23.4 (gnu/linux)


OK, current Version_2_6_10pre does the regression-fresh without error
when preceded by

(si::allocate 'contiguous 15000 t)
(si::allocate-sgc 'contiguous 15000 100000 10)

If you have a git repository checked out to this version, you can cd
there and do a 'git pull' instead of having to clone the repository all
over again.  'git status' will tell you what version is checked out.

I am suggesting leaving the hole size and log-maxpage-bound alone.

The default acl2 build without the extra contiguous allocation uses
about 12000 pages. (apparently all that compiled code in TMP1)  This is
sufficient for all books except out friend elementary-bounders, where
the last load occurs over the 2Gb limit.  One can probably do with a
smaller allocation (but greater than the default number as reported by
room), but 15000 works.

Please let me know what you think, or if this requirement makes things
'too fiddly'.  As I said earlier, appending "-mcmodel=large" to
compiler::*cc* with no allocation also works at ut, but has the problem
that it does not yet compile gcl itself.

Meanwhile, all autobuilds look fine but ia64 at the moment, so we're
getting close to release, hopefully a few days.

Take care,

Matt Kaufmann <address@hidden> writes:

> Hi, Camm --
> I'm happy to "hold off" -- but I'm not sure what activity I'm
> postponing.  Are you referring to my offer to run a new regression,
> made within the last hour (below)?  If so, there were a few questions
> in there to which I'd first like the answer.
>   Hi, Camm --
>   Thanks for the progress.
>   I think you're suggesting that I keep the changes in
>   /projects/acl2/camm-09-19/from-matt-09-19/acl2-init.lisp (as compared
>   to /projects/acl2/camm-09-19/acl2-init.lisp), so that ACL2 no longer
>   messes with the hole size, but that I do NOT keep the change in
>   /projects/acl2/camm-09-19/from-matt-09-19/acl2.lisp (as compared to
>   /projects/acl2/camm-09-19/acl2.lisp), which was calling
>   si::set-log-maxpage-bound.  Do I have that right?  Have you run a
>   successful regression at UT CS with those changes?  If not, I don't
>   mind doing so for you, using an updated GCL.  Should I obtain/install
>   GCL as you suggested before, like this?
>   git clone git://git.sv.gnu.org/gcl.git
>   cd gcl
>   git checkout Version_2_6_10pre
>   cd gcl
>   ./configure && make 
>   Thanks --
>   -- Matt
> Thanks --
> -- Matt
>    From: Camm Maguire <address@hidden>
>    Date: Sat, 21 Sep 2013 08:59:28 -0400
>    Greetings!  Just a quick note that while my regression built atop gcl
>    with --enable-debug completed without failure, an issue with C
>    optimization turned on has just presented itself, so if you want to hold
>    off for a little, I'll let you know when this is resolved.
>    Take care,
>    -- 
>    Camm Maguire                                           address@hidden
>    ==========================================================================
>    "The earth is but one country, and mankind its citizens."  --  Baha'u'llah

Camm Maguire                                        address@hidden
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]