Hello –

 

I can confirm that we are unable to allocate more than ~28 GB of RAM for an APBS calculation and that this problem persists in APBS 1.4.  I’ve asked our team to look into a fix but don’t have an estimate for when this fix will be complete.

 

Sorry,

 

__________________________________________________
Nathan Baker
Pacific Northwest National Laboratory
Tel:  +1-509-375-3997

http://nabaker.me

 

From: Kelly Tran [mailto:trankellyn@gmail.com]
Sent: Tuesday, December 11, 2012 2:05 PM
To: apbs-users@googlegroups.com
Cc: M.V.Vulovic@lumc.nl; berenger@riken.jp; apbs-users@lists.sourceforge.net; Baker, Nathan
Subject: Re: [apbs-users] maximum memory limit inputgen

 

Hello,

 

Did you find the bug and is there a fix? I have a similar problem w/ APBS 1.3 but I have not tried upgrading to 1.4 yet.

 

 

Thank you,

 

Kelly

On Tuesday, August 7, 2012 7:13:58 PM UTC-4, Baker, Nathan wrote:

Hello --

Can you send the input file you tried to use?  That will help us determine where the bug is.

Thanks,

__________________________________________________
Nathan Baker
Pacific Northwest National Laboratory
Tel:  +1-509-375-3997
http://nabaker.me


-----Original Message-----
From: M.V.V...@lumc.nl [mailto:M.V.V...@lumc.nl]
Sent: Tuesday, August 07, 2012 3:54 PM
To: bere...@riken.jp; apbs-...@lists.sourceforge.net
Cc: apbs-...@googlegroups.com
Subject: Re: [apbs-users] maximum memory limit inputgen

Hi all,
ulimit -a  shows that the memory is 'unlimited' and that is not a problem.
However, the failed amount of memory is printed as a %u. If you print it as a %lu you'll see it really tried claiming more then the amount of available memory (compare the lines 2 and 3 in the bellow error message):
1. FATAL ERROR: Unable to allocate enough memory for problem size.
2. FATAL ERROR: You requested 2203582984 bytes of RAM 3. FATAL ERROR: You requested 18446744067323200008 bytes of RAM.
4. FATAL ERROR: Check that your system has that much memory.
5. VASSERT: ASSERTION FAILURE!  filename vmem.c, line 255, (ram != ((void *)0))

When it is printed as %u, it requires 2GB, with %lu it corresponds to ~1.7x10^11 GB.  Is this true?  When I use psize.py to estimate the amount of memory needed I got a requirement which is quite lower (see attachment). It requires ~33GB RAM which should not be a problem.
Do you know where does this discrepancy come from (33GB vs 1.7x10^11) ?  
Many thanks,
Milos

N.B. In the attached file one can see that the total charge for this large Hemoglobin molecule is -1284e which is off course not correct. It comes directly from the PDB model.  Could you advise me what would be the best way to make this protein neutral? Thank you very much!!
 


________________________________________
From: Francois Berenger [bere...@riken.jp]
Sent: Monday, August 06, 2012 3:33 AM
To: apbs-...@lists.sourceforge.net
Subject: Re: [apbs-users] maximum memory limit inputgen

Maybe look at the output of 'ulimit -a'
to check if some limits are too low for your use case.

On 08/06/2012 04:20 AM, M.V.V...@lumc.nl wrote:
> Dear,
> I tried as well with the new version 1.4 but still get the same message.
> I installed first the binary /APBS-1.4-linux-x86_64/ (which btw
> doesn't have anymore inputgen.py?) I compiled and installed also the
> source version /APBS-1.4-source.
> /Still the same message.
> Have I downloaded the wrong files?
> So at the moment, I set the --gmemceil=2000 of inputgen.py which
> corresponds to the limit ~2G RAM for 32 build. Then I got the input
> file HB.in which suggests mg-para calculations with pdime 4 4 4. Does
> this mean I need to run apbs 4*4*4=64 times with async from 0:63 and
> then to merge it afterwards with mergedx2?
> Do you have another solution?
> With many thanks,
> Milos
>
> ----------------------------------------------------------------------
> --
> *From:* Baker, Nathan [Nathan...@pnnl.gov]
> *Sent:* Saturday, August 04, 2012 2:55 AM
> *To:* Monson, Kyle E; Vulovic, M.V. (MCB);
> apbs-...@lists.sourceforge.net
> *Cc:* apbs-...@googlegroups.com
> *Subject:* RE: maximum memory limit inputgen
>
> Hello -
>
> We have a 64-bit build for APBS 1.4.  Have you tried that?
>
> Thanks,
>
> __________________________________________________
> *Nathan Baker*
> Pacific Northwest National Laboratory
> Tel:  +1-509-375-3997
> http://nabaker.me <http://nabaker.me/>
>
> *From:*Monson, Kyle E
> *Sent:* Friday, August 03, 2012 9:35 AM
> *To:* Baker, Nathan; M.V.V...@lumc.nl;
> apbs-...@lists.sourceforge.net
> *Cc:* apbs-...@googlegroups.com
> *Subject:* RE: maximum memory limit inputgen
>
> This can only be overcome with a 64-bit build of APBS. A 32-bit
> executable can only have a heap of about 2GB as you have seen.
>
> I'm not sure what our plans are for a 64-bit build.
>
> *From:*Baker, Nathan
> *Sent:* August 03, 2012 06:32
> *To:* M.V.V...@lumc.nl; apbs-...@lists.sourceforge.net
> *Cc:* apbs-...@googlegroups.com
> *Subject:* Re: [apbs-users] maximum memory limit inputgen
>
> Hello -
>
> Do you have the same problem with the new APBS 1.4?
>
> Thanks,
>
> __________________________________________________
> *Nathan Baker*
> Pacific Northwest National Laboratory
> Tel:  +1-509-375-3997
> http://nabaker.me <http://nabaker.me/>
>
> *From:*M.V.V...@lumc.nl <mailto:M.V.V...@lumc.nl>
> [mailto:M.V.V...@lumc.nl]
> *Sent:* Friday, August 03, 2012 5:54 AM
> *To:* apbs-...@lists.sourceforge.net
> <mailto:apbs-...@lists.sourceforge.net>
> *Cc:* apbs-...@googlegroups.com <mailto:apbs-...@googlegroups.com>
> *Subject:* [apbs-users] maximum memory limit inputgen
>
> Dear all,
>
> I was trying to upped the maximum allowable memory used by APBS. I
> would like to calculate the potential distribution for a larger
> Hemoglobin molecule. Machine on which I am running the program has 256GB RAM memory.
>
> Unfortunately, I always get the same message:
>
> --------------------------------------
>
> Unable to allocate memory!
>
> (This often means you don't have enough memory available for this
> calculation.) FATAL ERROR: Unable to allocate enough memory for
> problem size.
>
> FATAL ERROR: You requested 2203582984 bytes of RAM.
>
> FATAL ERROR: Check that your system has that much memory.
>
> VASSERT: ASSERTION FAILURE!  filename vmem.c, line 252, (ram != ((void
> *)0))
>
> ----------------------------------------------------------------------
>
> This corresponds to 'only' ~2GB RAM. Could you help me to overcome
> this limitation?
>
> I first call inputgen.py that should allow me to define the maximum
> memory :~ python ~/soft/apbs-1.3-amd64/share/tools/manip/inputgen.py
> --gmemceil=100000 Hb.pqr
>
> And then as a second step I run apbs program on a newly generated  
> file Hb.in ~/soft/apbs-1.3-amd64/bin/apbs Hb.in
>
> And this still gives me the same error message above.
>
> Do you know what could be the problem? Am I doing something wrong?
>
> thanks a lot
>
> Milos
>
> Leiden University Medical Center
> Department of Molecular Cell Biology
> Section Electron Microscopy
> Postal Zone S1-P
> P.O.Box 9600
> 2300 RC Leiden
> The Netherlands
> Visiting address: Einthovenweg 20
> 2333 ZC Leiden
> E-mail: vul...@lumc.nl <mailto:vul...@lumc.nl>
> Tel:  (+31) 71 526 9265
> Fax: (+31) 71 526 8270
>
>
>
> ----------------------------------------------------------------------
> --------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond.
> Discussions will include endpoint security, mobile security and the
> latest in malware threats.
> http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>
>
>
> _______________________________________________
> apbs-users mailing list
> apbs-...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/apbs-users
>


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
apbs-users mailing list
apbs-...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/apbs-users

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
apbs-users mailing list
apbs-...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/apbs-users