Re: memory exhausted

Re: memory exhausted

[Rik van Riel]

[mailing list address corrected ... won't people ever learn to read ?]

On Wed, 24 Apr 2002, Vivian Wang wrote:

> I try to sort my 11 GB file, but I got message about memory exhausted.
> I used the command like this:
> sort -u file1 -o file2
> Is this correct?

Yes, sort only has a maximum of 3 GB of virtual address space so
it will never be able to load the whole 11 GB file into memory.

> What I should do?

You could either write your own sort program that doesn't need
to have the whole file loaded or you could use a 64 bit machine
with at least 11 GB of available virtual memory, probably the
double...

regards,

Rik
-- 
Bravely reimplemented by the knights who say "NIH".

http://www.surriel.com/		http://distro.conectiva.com/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: memory exhausted

[William Lee Irwin III]

On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> [mailing list address corrected ... won't people ever learn to read ?]

On Wed, 24 Apr 2002, Vivian Wang wrote:
>> I try to sort my 11 GB file, but I got message about memory exhausted.
>> I used the command like this:
>> sort -u file1 -o file2
>> Is this correct?

On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> Yes, sort only has a maximum of 3 GB of virtual address space so
> it will never be able to load the whole 11 GB file into memory.

This is larger than the virtual address space of i386 machines,
but not larger than the physical address space. In principle, an
executive taking advantage of 36-bit physical addressing extensions and
performing its own memory management on the bare metal could perform an
in-core sort on a 36-bit physical addressing -capable 32-bit machine,
e.g. i386-style PAE/highmem machines and some 32-bit MIPS machines. A
kernel module could also in principle take advantage of the kernel's
low-level memory management facilities to perform such an in-core sort.
While possible, this is absolutely not recommended.


On Wed, 24 Apr 2002, Vivian Wang wrote:
>> What I should do?

On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> You could either write your own sort program that doesn't need
> to have the whole file loaded or you could use a 64 bit machine
> with at least 11 GB of available virtual memory, probably the
> double...
> regards,
> Rik

It's doubtful the above "solutions" I mentioned above are practical for
your purposes unless you are under the most extreme duress and have
access to uncommon hardware. I suggest polyphase merge sorting or any
of the various algorithms recommended in Donald E. Knuth's "The Art of
Computer Programming", specifically its chapter on external sorting,
which I'm willing to discuss and assist in implementations of off-list.


Cheers,
Bill
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: memory exhausted

[msimons]

> On Wed, 24 Apr 2002, Vivian Wang wrote:
> >> I try to sort my 11 GB file, but I got message about memory exhausted.
> >> sort -u file1 -o file2
> >> What I should do?

Vivian,

  If you are using the GNU version of sort it does not even try to load 
the input set into memory, it does a splintering and merge multiple merge 
sorts of the input files.  The sort operation will be very disk I/O bound...

  You must set -T (or TMPDIR) to point at a filesystem with enough disk 
space to store the temporary files if your /tmp doesn't have enough room.
You didn't paste the exact text of the error message so I didn't check 
to see what error is generated if /tmp fills up while working.


  It appears there is a restriction that sort must be able to load a few
of the longest lines into memory.

- What is the longest line in your input file?
  (run "wc -L input file")

  In my own testing of very big data files (with short line lengths)
I had no problems.  If that is a number larger than memory you may
need to find a way to chop your data into shorter lines...


  Lastly, sort splits the files into 500 KiB byte chunks on the first
pass.  This will create about 22000 files in your TMPDIR location on 
first pass, if you are using a filesystem with any sort of number of
files per directory limitation you could be having a problem there.


  Check those things and if you are still having a problem, send the
error messages you see, and some more information (like longest line).

    Good Luck,
      Mike

ps:
  I think this question may be off topic, since it's a userspace not kernel
problem... so if anyone complains you might want to take it off channel
or to some GNU textutils related mailing list.


On Wed, Apr 24, 2002 at 07:57:53PM -0700, William Lee Irwin III wrote:
> On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> On Wed, 24 Apr 2002, Vivian Wang wrote:
> >> I try to sort my 11 GB file, but I got message about memory exhausted.
> >> I used the command like this:
> >> sort -u file1 -o file2
> >> Is this correct?
> 
> On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> > Yes, sort only has a maximum of 3 GB of virtual address space so
> > it will never be able to load the whole 11 GB file into memory.
[...]
> On Wed, 24 Apr 2002, Vivian Wang wrote:
> >> What I should do?
> 
> On Wed, Apr 24, 2002 at 11:19:50PM -0300, Rik van Riel wrote:
> > You could either write your own sort program that doesn't need
> > to have the whole file loaded or you could use a 64 bit machine
> > with at least 11 GB of available virtual memory, probably the
> > double...
> > regards,
> > Rik
> 
> It's doubtful the above "solutions" I mentioned above are practical for
> your purposes unless you are under the most extreme duress and have
> access to uncommon hardware. I suggest polyphase merge sorting or any
> of the various algorithms recommended in Donald E. Knuth's "The Art of
> Computer Programming", specifically its chapter on external sorting,
> which I'm willing to discuss and assist in implementations of off-list.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: memory exhausted

[H. Peter Anvin]

William Lee Irwin III wrote:
> 
> This is larger than the virtual address space of i386 machines,
> but not larger than the physical address space. In principle, an
> executive taking advantage of 36-bit physical addressing extensions and
> performing its own memory management on the bare metal could perform an
> in-core sort on a 36-bit physical addressing -capable 32-bit machine,
> e.g. i386-style PAE/highmem machines and some 32-bit MIPS machines. A
> kernel module could also in principle take advantage of the kernel's
> low-level memory management facilities to perform such an in-core sort.
> While possible, this is absolutely not recommended.
> 

Good God, I hope x86-64 catches on soon and kills off this PAE silliness...

	-hpa


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: memory exhausted

[William Lee Irwin III]

William Lee Irwin III wrote:
>> This is larger than the virtual address space of i386 machines,
>> but not larger than the physical address space. In principle, an
>> executive taking advantage of 36-bit physical addressing extensions and
>> performing its own memory management on the bare metal could perform an
>> in-core sort on a 36-bit physical addressing -capable 32-bit machine,
>> e.g. i386-style PAE/highmem machines and some 32-bit MIPS machines. A
>> kernel module could also in principle take advantage of the kernel's
>> low-level memory management facilities to perform such an in-core sort.
>> While possible, this is absolutely not recommended.


On Sat, Apr 27, 2002 at 12:30:49PM -0700, H. Peter Anvin wrote:
> Good God, I hope x86-64 catches on soon and kills off this PAE silliness...
> 	-hpa


Taunting me, eh? =)

Well, I did say "absolutely not recommended". 64-bit hardware of
whatever kind is without question a more appropriate solution to these
kinds of issues than such shenanigans anyway, and at this point I'm
more or less sorry I brought that up. And I'll leave the discussion
of what specific lines of hardware are most suitable for other fora. =)


Cheers,
Bill
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: memory exhausted

[H. Peter Anvin]

William Lee Irwin III wrote:
> 
>>Good God, I hope x86-64 catches on soon and kills off this PAE silliness...
> 
> Well, I did say "absolutely not recommended". 64-bit hardware of
> whatever kind is without question a more appropriate solution to these
> kinds of issues than such shenanigans anyway, and at this point I'm
> more or less sorry I brought that up. And I'll leave the discussion
> of what specific lines of hardware are most suitable for other fora. =)
> 

Oh, no, I didn't mean to pick on you.  It was more of a general lament. 
  I hope x86-64 can make 64-bit computing widespread.

	-hpa


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

cpp: memory exhausted

[Matthew Studley]

Hi all

I am compiling some code using gcc.  I have compiled the same code before,
on the same platform, with the same compiler, on the same ... nothing's
changed!

But now I get the message:

cpp: memory exhausted

I have rebooted, physically switched off and on - still the same problem!

help!

please?

Matt
==============================================
Matthew Studley

Intelligent Autonomous Systems Laboratory

Faculty of Computing, Engineering
        and Mathematical Sciences
Coldharbour Lane, Frenchay
BS16 1QY
United Kingdom

tel : 07092 306548
fax : 07092 306548

Re: memory exhausted

[Matthew Studley]

Well I clearly have a brain like a seive - I now remember that I've had this
problem before.  To solve it I deleted all object files and recompiled the
entire application.  No problem.  Your guesses why this happens would be
very interesting!

regards

matt
<phew> panic over!

----- Original Message -----
From: Matthew Studley <matthew.studley@uwe.ac.uk>
To: <linux-c-programming@vger.kernel.org>
Sent: Tuesday, November 25, 2003 2:51 PM
Subject: cpp: memory exhausted


> Hi all
>
> I am compiling some code using gcc.  I have compiled the same code before,
> on the same platform, with the same compiler, on the same ... nothing's
> changed!
>
> But now I get the message:
>
> cpp: memory exhausted
>
> I have rebooted, physically switched off and on - still the same problem!
>
> help!
>
> please?
>
> Matt
> ==============================================
> Matthew Studley
>
> Intelligent Autonomous Systems Laboratory
>
> Faculty of Computing, Engineering
>         and Mathematical Sciences
> Coldharbour Lane, Frenchay
> BS16 1QY
> United Kingdom
>
> tel : 07092 306548
> fax : 07092 306548
>
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe
linux-c-programming" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
> This incoming email to UWE has been independently scanned for viruses and
any virus detected has been removed using McAfee anti-virus software
>
>
>