From: Jim Phillips (jim_at_ks.uiuc.edu)
Date: Tue Jun 07 2011 - 18:02:21 CDT
Thanks for reporting this. Since this is a 32-bit build you'll run out of
address space before you run out of memory. That said, the thread stacks
are supposed to be 128K on both Linux and Windows, and you'll get
13*13*13+1 = 2198 of them, so that would be at most 300 MB additional.
We need to fix Charm++ to report the underlying error condition to see
what's actually wrong. You might try the Win64-MPI build (it needs
mpiexec from HPC Server).
-Jim
On Tue, 7 Jun 2011, Ajasja Ljubeti~M wrote:
> Hi,
>
> I just manage to produce the same error running the stmv benchmark on win32.
> I have 8 GB of ram, on linux the stmv needs about 800 MB.
>
> here are the important bits from the log file:
>
>
> Info: NAMD 2.8 for Win32-multicore
> Info: Based on Charm++/Converse 60303 for multicore-win32
> Info: Built Sat May 28 11:36:37 CDT 2011 by jcphill on honor
> Info: Running on 8 processors, 1 nodes, 1 physical nodes.
> Info: CPU topology information available.
> Info: Charm++/Converse parallel runtime startup completed at 0.252 s
> .
> .
> .
> Info: Startup phase 8 took 0.131 s, 774.531 MB of memory in use
> Program finished.
>
>
> Best regards,
> Ajasja
>
> On Fri, Jun 3, 2011 at 18:42, Jim Phillips <jim_at_ks.uiuc.edu> wrote:
>
>>
>> Does this happen for any simulation or just this one? The rest of the log
>> might provide some useful information for diagnosing this error.
>>
>> -Jim
>>
>>
>> On Fri, 3 Jun 2011, Patriche Simona wrote:
>>
>> Dear NAMD users,
>>>
>>> When I tried to perform the minimization step in my molecular dynamics
>>> simulation, I received this error: CthCreate failed to create fiber.
>>>
>>> It is possible to receive this error because I have not enough resources
>>> on my
>>> system?
>>>
>>> Thank you in advance for any sugestions,
>>>
>>> Simona Patriche
>>>
>>
>>
>
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:24 CST