[Gcl-devel] Re: 32bit GCL

[Matt Kaufmann]

Hi, Camm --

Thanks.  I don't mind missing a Debian release.  If you become in more
of a rush, let me know.

You asked:

>> BTW, does SGC speed things up for you these days?  I know how to
>> extend this to mac and windows, but have not done so yet.

I did a sort of modified ACL2 regression run on a 24-processor
machine, using make -j 24.  The results aren't perfectly comparable
(e.g., the first had to suffer migrating some jobs off the machine),
but I think they're comparable enough.

SGC off:
21514.628u 1100.780s 27:11.97 1385.7%   0+0k 218184+1907232io 0pf+0w

SGC on:
20277.963u 1266.835s 26:46.05 1341.4%   0+0k 229800+1901360io 0pf+0w

This was on 64-bit Linux:

oliphaunt-0:~> uname -a
Linux oliphaunt-0.cs.utexas.edu 2.6.24.6+desktop10+25.63 #1 SMP Thu Nov 5 
15:16:15 CST 2009 x86_64 GNU/Linux
oliphaunt-0:~> 

-- Matt
   Cc: address@hidden, address@hidden, address@hidden
   From: Camm Maguire <address@hidden>
   Date: Mon, 09 Aug 2010 15:09:41 -0400
   X-SpamAssassin-Status: No, hits=-2.6 required=5.0
   X-UTCS-Spam-Status: No, hits=-252 required=165

   Greetings!

   Matt Kaufmann <address@hidden> writes:

   > Hi, Camm  --
   >
   > Thanks for the fix!  At one time, that same form (setq
   > si::*multiply-stacks* 2), also failed on Windows.  Does it work on
   > Windows now?
   >

   Works for me under wine, so in all likelihood.

   > I haven't forgotten to get you a fix for the issue of moving .cert
   > files in ACL2.  I have a fix, but I'm considering just doing a new
   > ACL2 release.  Let me know if you have any particular timetable for
   > getting the fix.

   Thanks so much.  I'm in no rush, but there is the question of a
   forthcoming Debian release which we might miss.  If this is a
   consideration I can try to get more info regarding the timetable.

   BTW, does SGC speed things up for you these days?  I know how to
   extend this to mac and windows, but have not done so yet.

   Take care,



   >
   > Thanks --
   > -- Matt
   >    Cc: address@hidden, address@hidden, address@hidden
   >    From: Camm Maguire <address@hidden>
   >    Date: Sun, 08 Aug 2010 22:53:49 -0400
   >    X-MagicMail-UUID: 5af74938-a361-11df-b7c3-000c29c6406d
   >    X-SpamAssassin-Status: No, hits=-2.6 required=5.0
   >    X-UTCS-Spam-Status: No, hits=-282 required=165
   >
   >    Greetings!
   >
   >    Matt Kaufmann <address@hidden> writes:
   >
   >    > Hi, Camm --
   >    >
   >    > I've just grabbed the latest cvs version of GCL 2.6.8pre and built
   >    > ACL2 with it, on my Intel Mac running Mac OS 10.6.4.  The build
   >    > succeeded, as did a run of the ACL2 regression suite (using a
   >    > development copy of ACL2, not much different from ACL2 4.0) -- well,
   >    > almost.  (But almost is still great -- thanks!)  Here's the story.
   >    >
   >    > I first removed the progn form near the end of source file init.lisp
   >    > that contains the form
   >    >
   >    > (setq si::*multiply-stacks* 2)
   >    >
   >    > because that form caused an error (though maybe the build would have
   >    > succeeded; I'm not sure).  Here's a simple log from my Mac, not
   >    > involving ACL2:
   >    >
   >
   >    Thanks for spotting this -- looks like its very old.  In any case,
   >    should be fixed now.  Please let me know if problems persist.
   >
   >    Take care,
   >
   >    >   ~$ /Users/kaufmann/lisps/gcl/gcl-2.6.8pre/unixport/saved_gcl
   >    >   GCL (GNU Common Lisp)  2.6.8 CLtL1    Aug  6 2010 11:51:51
   >    >   Source License: LGPL(gcl,gmp), GPL(unexec,bfd,xgcl)
   >    >   Binary License:  GPL due to GPL'ed components: (READLINE UNEXEC)
   >    >   Modifications of this banner must retain notice of a compatible 
license
   >    >   Dedicated to the memory of W. Schelter
   >    >
   >    >   Use (help) to get some basic information on how to use GCL.
   >    >   Temporary directory for compiler files set to 
/var/folders/6S/6ScqCriVFE8aH0vpDjfNK++++TI/-Tmp-/
   >    >
   >    >   >(setq si::*multiply-stacks* 2)
   >    >
   >    >   2
   >    >
   >    >   >
   >    >   Error: Caught fatal error [memory may be damaged]
   >    >   Fast links are on: do (si::use-fast-links nil) for debugging
   >    >   Error signalled by an anonymous function.
   >    >
   >    >   Error: 4 is an illegal ihs index.
   >    >   Fast links are on: do (si::use-fast-links nil) for debugging
   >    >   Error signalled by SYSTEM:UNIVERSAL-ERROR-HANDLER.
   >    >   Broken at SYSTEM:SCH-FRS-BASE.
   >    >   >>>:b
   >    >   Backtrace: system:universal-error-handler > SYSTEM:SCH-FRS-BASE
   >    >   NIL
   >    >   >>>
   >    >
   >    > Perhaps there's a different way I should increase stack size?  One
   >    > regression test failed, namely books/serialize/serialize-tests.lisp,
   >    > with "Invocation history stack overflow.".  I'm guessing it's because
   >    > I didn't double the stack size, though I'm not sure.
   >    >
   >    > Thanks --
   >    > -- Matt
   >    >    Cc: address@hidden, address@hidden, Robert Krug <address@hidden>
   >    >    From: Camm Maguire <address@hidden>
   >    >    Date: Fri, 06 Aug 2010 15:12:15 -0400
   >    >    X-SpamAssassin-Status: No, hits=-2.6 required=5.0
   >    >    X-UTCS-Spam-Status: No, hits=-252 required=165
   >    >
   >    >    Greetings!  Wonderful to hear from you too!
   >    >
   >    >    "Warren A. Hunt Jr." <address@hidden> writes:
   >    >
   >    >    > Hi Camm,
   >    >    >
   >    >    > It's good to hear that you are back at work on GCL.
   >    >    >
   >    >    >   It appears we're getting close to a gcl release.  You might 
recall out
   >    >    >   [typo?; probably should be be "our"]
   >    >    >   having used static builds in the past to enable 32bit linux 
machines
   >    >    >   to use up to 3gig memory as opposed to the usual 1gig limit 
(imposed
   >    >    >   by the load address of shared libraries.)  Warren told me at 
one time
   >    >    >   that this was useful in getting the most out of 32bit, 
especially as
   >    >    >   64bit comes with its own overhead of bigger pointers.  In 
addition,
   >    >    >   the binary of course is completely portable.  Is this 
important to
   >    >    >   support?  There appear to have been some libc developments 
which will
   >    >    >   have to be worked around to get it working now.
   >    >    >
   >    >    > I believe there will be interest in 32-bit computing
   >    >    > for many years to come; I believe the embedded world of
   >    >    > computing will become almost entirely based on 32-bit
   >    >    > ARM and X86 platforms.  The number of embedded systems
   >    >    > (phones, tablets, set-top boxes, utility meters, etc.)
   >    >    > far exceeds the number of general-purpose computing
   >    >    > systems.  For instance, the device that measures the
   >    >    > electric power used in my home is an embedded Linux
   >    >    > system running on a Pico ITX board with a X86
   >    >    > processor.  So, for embedded computing, I think getting
   >    >    > what one can from the 32-bit X86 architecture is
   >    >    > valuable.
   >    >    >
   >    >    > Some years ago, Boyer and I didn't have ready access to
   >    >    > 64-bit computers so we were motivated to get as much
   >    >    > out of our 32-bit computers as possible.  However, now
   >    >    > even my laptop contains a 64-bit (Intel Core 2 Duo)
   >    >    > microprocessor with 8 GBytes RAM, so now I always work
   >    >    > on a 64-bit machine.  Therefore, I don't expect to be a
   >    >    > future customer of 32-bit GCL for general-purpose
   >    >    > computing.
   >    >    >
   >    >
   >    >    Good thing then I guess that I put in the extra work to get 64bit
   >    >    working on the mac.  There is a special configure option needed as
   >    >    explained in README.macosx.  The standard configure tools detect 
the
   >    >    box as 32bit by default.
   >    >
   >    >    In any case, this mac stuff appears done now.  If anyone finds it
   >    >    useful enough to try it out and report back, that of course would 
be
   >    >    great.  No need of course unless it is of genuine use to some real 
mac
   >    >    user. 
   >    >
   >    >    Take care,
   >    >
   >    >    > Cheers,
   >    >    >
   >    >    > Warren
   >    >    >
   >    >    >
   >    >    >
   >    >    >
   >    >
   >    >    -- 
   >    >    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
   >
   >
   >
   >

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