Dear NM users,
I've been encountering a problem when using NONMEM 7.3 and the file passing
interface for parallel computing.
The run I'm trying to get going consists of 5 problems within a single NONMEM
run.
Here is a short extract of the script:
$PROBLEM Fit cohort 1
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.EQ.1) REWIND
...
$ESTIMATION ... MSFO=run1.msf
---
$PROBLEM Post hoc predictions
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.NE.1) REWIND
$MSFI run1.msf
...
$ESTIMATION ... MAX=0
---
$PROBLEM Fit cohort 2
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.EQ.2) REWIND
...
$ESTIMATION ... MSFO=run1.msf
---
$PROBLEM Post hoc predictions
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.NE.2) REWIND
$MSFI run1.msf
...
$ESTIMATION ... MAX=0
---
$PROBLEM Fit all data
$INPUT ...
$DATA data.csv IGNORE=@ REWIND
...
$ESTIMATION ...
The script works perfectly fine without the parallel computing option. When
using the fpi I get the following error message:
At line 169 of file <unit = 26, file = ´FILE26´>
Fortran runtime error: End of File
The run consistently fails when initiating problem 5 (i.e. the initial OFV
evaluation).
I've searched the NONMEM guides, and tried looking for some information online
on gfortran but I was not able to identify the problem.
Hopefully someone on this forum can shed some light on this behavior.
Warm regards,
Pieter Colin
Department of Anesthesiology
University Medical Center Groningen
Problem with fpi in NONMEM 7.3
Pieter:
Generally, using MPI parallelization method is preferred. The FILE26 is one of
the internal buffer files used by default to save on memory usage. As
now-a-days computers have lots of memory, you can add the following option
switch to your command line (which works with nonmem 7.3 and above):
-maxlim=3
and this will request NMTRAN to size all internal memory so no buffer files are
needed. This should prevent the error you are seeing.
Nonetheless, the error should not have occurred even when using FPI, and
suggests that improper transfer of file buffer data between manager and workers
is occurring. If you are able, please send me an example control stream and
data file, so I may determine if something needs to be corrected. Also, please
let me know what version of NONMEM you are using.
Robert J. Bauer, Ph.D.
Senior Director
Pharmacometrics R&D
ICON Early Phase
820 W. Diamond Avenue
Suite 100
Gaithersburg, MD 20878
Office: (215) 616-6428
Mobile: (925) 286-0769
[email protected]<mailto:[email protected]>
http://www.iconplc.com/
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Pieter Colin
Sent: Thursday, January 04, 2018 2:09 AM
To: '[email protected]'
Subject: [NMusers] Problem with fpi in NONMEM 7.3
Dear NM users,
I've been encountering a problem when using NONMEM 7.3 and the file passing
interface for parallel computing.
The run I'm trying to get going consists of 5 problems within a single NONMEM
run.
Here is a short extract of the script:
$PROBLEM Fit cohort 1
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.EQ.1) REWIND
...
$ESTIMATION ... MSFO=run1.msf
---
$PROBLEM Post hoc predictions
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.NE.1) REWIND
$MSFI run1.msf
...
$ESTIMATION ... MAX=0
---
$PROBLEM Fit cohort 2
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.EQ.2) REWIND
...
$ESTIMATION ... MSFO=run1.msf
---
$PROBLEM Post hoc predictions
$INPUT ...
$DATA data.csv IGNORE=@ IGNORE(COHORT.NE.2) REWIND
$MSFI run1.msf
...
$ESTIMATION ... MAX=0
---
$PROBLEM Fit all data
$INPUT ...
$DATA data.csv IGNORE=@ REWIND
...
$ESTIMATION ...
The script works perfectly fine without the parallel computing option. When
using the fpi I get the following error message:
At line 169 of file <unit = 26, file = ´FILE26´>
Fortran runtime error: End of File
The run consistently fails when initiating problem 5 (i.e. the initial OFV
evaluation).
I've searched the NONMEM guides, and tried looking for some information online
on gfortran but I was not able to identify the problem.
Hopefully someone on this forum can shed some light on this behavior.
Warm regards,
Pieter Colin
Department of Anesthesiology
University Medical Center Groningen