RE: New versus old NONMEM
Dear Koen,
I agree with Nick that I cannot see any good reason to use NONMEM V anymore
for a new project. The FOCE+I algorithm in NONMEM VI was in several cases a lot
more stable than FOCE+I (with the SLOW option) in NONMEM V.
However, if FOCE+I in NONMEM V ran fine, the results I got during re-analysis
of the same
datasets in NONMEM VI were either "identical" or very similar to the results in
NONMEM V.
I ran such a comparison over around 5-10 datasets and this was on different
compilers,
CPUs, and operating systems. Some complex models ran however more robust in
NONMEM VI than in NONMEM V.
The Uppsala group probably has more experience on this than anybody else, since
they
used NONMEM VI beta for several years in parallel to NONMEM V, as far as I know.
NONMEM VII of course offers you a series of EM-type algorithms that are not
available in NONMEM VI and that do not require an approximation to the
likelihood
equation. If you are not in a hurry, you might wait for NONMEM VIII which will
offer
significant benefits for parallelization, for example.
I would select the operating system according to the recommendations of NONMEM
VIII,
since installing a parallelized version will probably be a bit more tricky. I
guess ICON
will support Win 2003 / 2008 Server 64 bit and a couple of distributions of
Linux and
it would be great to hear from our colleagues at ICON about this. Please take
care to
use a 64 bit OS to be able to use more than 3 GB of RAM.
Hope this helps.
Best wishes
Juergen
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Nick Holford
Sent: Saturday, October 23, 2010 2:32 AM
To: [email protected]
Subject: Re: [NMusers] New versus old NONMEM
Koen,
What are the advantages of NONMEM7 with respect to older versions like V and VI?
Automatic computation of eta and epsilon shrinkage.
The default objective function display format is easier for humans to read
(this can easily be implemented in NONMEM VI by changing the source code).
Is it still "acceptable" to use versions V and VI nowadays?
I consider NONMEM V is obsolete and people still using it either are using an
unlicensed version or have terrible IT support.
NONMEM VI release 2 is the most reliable version. There are bugs in NONMEM 7
that are not present in NONMEM VI (e.g. see recent email from Paulo Denti).
In my experience NONMEM VI is generally faster and uses less memory than NONMEM
7 which is a big advantage on multi-core machines. I can only run 3 problems
with NONMEM 7at the same time but I can run 8 problems with NONMEM VI on the
same 8 core machine. The next version of NONMEM 7 is expected to fix this
memory problem but it remains to be seen if it faster.
Is there an "optimal" combination of NONMEM, compiler and operating system?
The fastest compiler I know of is Intel Visual Fortran for both NONMEM VI and
NONMEM 7. There are occasionally differences in results between compilers but I
don't know of any systemic study that would favour one compiler over another in
terms of accuracy and consistency of results. I am not aware that the OS makes
any difference to the way that NONMEM runs if the same compiler and hardware
are used. Some virtual machine hosts may be slow down runs e.g. I find that
intensive disk activity e.g. creating simulation table files with Windows 2003
running as a virtual machine on top of Xen server is much slower than a native
Windows 2003 installation.
The main thing is to use the fastest CPU you can find!
Nick
On 23/10/2010 4:00 a.m., Jolling, Koen (Wavre) wrote:
Dear,
I would greatly appreciate your opinion on the following questions:
What are the advantages of NONMEM7 with respect to older versions like V and VI?
Is it still "acceptable" to use versions V and VI nowadays?
Is there an "optimal" combination of NONMEM, compiler and operating system?
Thanks,
Koen
Information in this email and any attachments is confidential and intended
solely for the use of the individual(s) to whom it is addressed or otherwise
directed. Please note that any views or opinions presented in this email are
solely those of the author and do not necessarily represent those of the
Company. Finally, the recipient should check this email and any attachments for
the presence of viruses. The Company accepts no liability for any damage caused
by any virus transmitted by this email. All SGS services are rendered in
accordance with the applicable SGS conditions of service available on request
and accessible at http://www.sgs.com/terms_and_conditions.htm
--
Nick Holford, Professor Clinical Pharmacology
Dept Pharmacology & Clinical Pharmacology
University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand
tel:+64(9)923-6730 fax:+64(9)373-7090 mobile:+64(21)46 23 53
email: [email protected]<mailto:[email protected]>
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford