[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Gcl-devel] Re: CONS Problem
[Gcl-devel] Re: CONS Problem
04 Jan 2005 12:44:40 -0500
Gnus/5.09 (Gnus v5.9.0) Emacs/21.2
Greetings, and thanks for the feedback! I'm assuming you are
referring to GCL's three word cons below. If so, which would you feel
has higher priority, a two word cons, or unlimited heap?
"Warren A. Hunt Jr." <address@hidden> writes:
> Hi Camm,
> I consider this very important. I sometimes use OpenMCL, even though
> it is slower, because it is more frugal in its memory usage and it
> doesn't seem to suffer from having to identify types of space that
> require specific allocation.
> To: address@hidden
> Cc: address@hidden, address@hidden,
> address@hidden, address@hidden
> Subject: Re: CONS Problem
> From: Camm Maguire <address@hidden>
> Date: 04 Jan 2005 10:15:15 -0500
> Greetings! How would you all rate the importance of removing fixed
> heap size bounds in GCL, allowing the image to expand until physical
> memory is exhausted? This would have to be done in a way which does
> not trigger any performance penalties as observed by Matt, which I
> believe are caused by the larger hole and relocatable block sizes that
> come perforce at present with any image configured with extended
> maxpages. (e.g. right now the hole is 1/10 maxpages). Rather these
> areas should grow dynamically with the heap as well.
> I think this could be done rather straightforwardly on Linux. I would
> have to ask our Windows and MacOSX people about it, as they emulate
> Unix's sbrk().
> BTW, to the fellow from rockwell -- we are working on unsigned long,
> long long, and unsigned long long support in the compiler right now in
> 2.7 (cvs head).
> Take care,
> Matt Kaufmann <address@hidden> writes:
> > Hi --
> > I believe that the memory limit we're discussing is completely determined
> > the underlying Lisp (GCL, Allegro CL, etc.). When you build ACL2
> > with "make"), the available memory is determined by how the underlying
> > executable was built.
> > Perhaps ACL2's installation instructions should clarify this point as
> > PLEASE NOTE: The available memory for ACL2 is determined by the
> > Common Lisp executable. If you need more memory, refer to your Common
> > instructions for building an executable.
> > I've added the above words for future releases to our development copy of
> > installation.html, section "Creating An Executable Image". Please let me
> > if it seems that I'm missing something here.
> > -- Matt
> > X-IronPort-MID: 429621954
> > X-SBRS: 3.5
> > X-BrightmailFiltered: true
> > X-Ironport-AV: i="3.88,96,1102312800";
> > d="scan'208"; a="429621954:sNHT21577160"
> > From: <address@hidden>
> > Date: Mon, 3 Jan 2005 10:19:05 -0600
> > X-MIMETrack: Serialize by Router on
> > 6.5.3|September 14, 2004) at 01/03/2005 10:26:40 AM,
> > Serialize complete at 01/03/2005 10:26:40 AM
> > Content-Type: multipart/alternative; boundary="=_alternative
> > Reply-To: address@hidden
> > Sender: address@hidden
> > X-Listprocessor-Version: 8.2.10/020311/17:52 -- ListProc(tm) by CREN
> > X-SpamAssassin-Status: No, hits=-2.5 required=5.0
> > X-UTCS-Spam-Status: No, hits=-290 required=180
> > This is a multipart message in MIME format.
> > --=_alternative 0059A39786256F7E_=
> > Content-Type: text/plain; charset="US-ASCII"
> > Matt, Bill, Warren:
> > Is there some way of making the memory space limit a compile-time
> > for some future ACL2 release? (A lot of other open source packages
> > the user to answer configuration questions before a compilation
> > using a configure script and then autoconf and automake.)
> > FYI, we have been routinely running out of CONS space here at Rockwell
> > since last summer. We have survived using a special
> > binary that Eric Smith built us. So the out-of-memory issue is a
> > more prevalent than one might think, and I suspect that more of us
> > run into this limit as the size of our projects (and the proof
> > capabilities of ACL2) slowly grow.
> > The text for the user query might read:
> > Do you want to allow ACL2 to use as much memory as possible [y/N]
> > By default, ACL2's maximum memory limit is set to a level
> > intended
> > to optimize overall performance. This limit is large
> > enough for almost
> > all applications. However, a very large proof set may
> > occasionally
> > need more memory. Say "n" here to use the default limit,
> > or "y" to use
> > as much memory as possible. If you aren't sure, answer
> > "n".
> > Hope this helps.
> > May the Force be with you,
> > -- Christian
> > Matt Kaufmann <address@hidden>
> > Sent by: address@hidden
> > 01/02/2005 12:01 PM
> > Please respond to
> > address@hidden
> > To
> > address@hidden, address@hidden
> > cc
> > address@hidden
> > Subject
> > Re: CONS Problem
> > Bill --
> > Are you using Linux or Solaris? If you are using Linux then I might
> > able to
> > build a larger image at UT (but let me know if you are using ACL2 2.9
> > ACL2
> > 2.9.1). It would be interesting to compare performance. Read on....
> > Warren --
> > I believe that in the past I've seen some performance degradation when
> > underlying GCL has enlarged memory. I suspect that only very few users
> > (specifically, you and Bill!) have run out of memory using ACL2 built
> > GCL
> > 2.6.5. Of course, users are welcome to build their own images, since
> > distribute the sources; the distributed image is only for convenience.
> > -- Matt
> > Date: Tue, 28 Dec 2004 10:51:48 -0600
> > From: "Warren A. Hunt Jr." <address@hidden>
> > CC: address@hidden, address@hidden
> > Hi Bill,
> > I don't use the default ACL2 binary because of the GCL on which the
> > default ACL2 is built. The storage limitations can be observed with
> > the (ROOM T) command (see below).
> > I always first build a GCL with the maximum pages set to 262,144,
> > which is about as large as can be built on a 32-bit system. This
> > provides a great deal more space.
> > Matt,
> > May I suggest that GCL-ACL2 systems be built with as large a memory
> > possible? The 1/2 GByte configurations are of lessening value.
> > Cheers,
> > Warren
> > ++++++
> > [hunt:~/] > acl2
> > GCL (GNU Common Lisp) 2.6.5 CLtL1 Aug 18 2004 10:07:03
> > Source License: LGPL(gcl,gmp), GPL(unexec,bfd)
> > Binary License: GPL due to GPL'ed components: (BFD 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.
> > ACL2 Version 2.9 built November 11, 2004 18:07:01.
> > Copyright (C) 2004 University of Texas at Austin
> > ACL2 comes with ABSOLUTELY NO WARRANTY. This is free software and
> > you
> > are welcome to redistribute it under certain conditions. For
> > details,
> > see the GNU General Public License.
> > Initialized with (INITIALIZE-ACL2 'INCLUDE-BOOK *ACL2-PASS-2-FILES*
> > NIL).
> > See the documentation topic note-2-9 for recent changes.
> > NOTE!! Proof trees are disabled in ACL2. To enable them in emacs,
> > look under the ACL2 source directory in interface/emacs/README.doc;
> > and, to turn on proof trees, execute :START-PROOF-TREE in the ACL2
> > command loop. Look in the ACL2 documentation under PROOF-TREE.
> > ACL2 Version 2.9. Level 1. Cbd "/v/filer2/hunt/s/intel-data/".
> > Type :help for help.
> > Type (good-bye) to quit completely out of ACL2.
> > ACL2 !>:q
> > Exiting the ACL2 read-eval-print loop. To re-enter, execute (LP).
> > ACL2>(room t)
> > 5141/5141 93.1% CONS RATIO LONG-FLOAT COMPLEX STRUCTURE
> > 43/28 45.8% FIXNUM SHORT-FLOAT CHARACTER
> > RANDOM-STATE READTABLE NIL
> > 207/209 99.1% SYMBOL STREAM
> > 1/2 37.2% PACKAGE
> > 227/228 0.3% ARRAY HASH-TABLE VECTOR
> > BIT-VECTOR PATHNAME CCLOSURE FAT-STRING
> > 167/172 91.0% STRING
> > 21/27 31.2% CFUN BIGNUM
> > 62/115 91.8% SFUN GFUN CFDATA SPICE NIL
> > 2599/2602 contiguous (2705 blocks)
> > 13036 hole
> > 5242 0.0% relocatable
> > 5869 pages for cells
> > 26746 total pages
> > 92978 pages available
> > 11348 pages in heap but not gc'd + pages needed for gc
> > marking
> > 131072 maximum pages
> > ACL2>
Camm Maguire address@hidden
"The earth is but one country, and mankind its citizens." -- Baha'u'llah
|[Prev in Thread]
||[Next in Thread]|
- [Gcl-devel] Re: CONS Problem,
Camm Maguire <=