Hi,
I know that this topic has come up from time to time, but I've not found
a definitive word on using NONMEM in virtualized environments. If I've
missed an old post here, please point me to the link I should have
found.
Specifically I have a few questions for people who have built
virtualized NONMEM clusters:
* As most of the readers of this list are probably aware, when
there is a very stiff problem, it is possible to get different results
on different CPU/OS/compiler combinations. When a virtualized system
(where the OS and compiler are held fixed) executes NONMEM on the same
CPU I would assume that identical results would come out. Do people
have experience with stiff problems on virtualized systems with
different underlying physical CPUs? Does it give the same or different
answers for such models?
* I know that current virtualization systems are pretty good about
not getting in the way very much for raw CPU power and IO/sec. Do
people have an idea on identical hardware how much the virtualization
system adds to the time to complete a NONMEM run? Hopefully there can
be a few answers here with different virtualization platforms (VMware,
KVM, Xen, ...).
* I've heard that the next version of NONMEM will support
parallelization within a run. I realize that people won't have
experience with this, but I was wondering if there are any hints of
about how well it will scale? Will it scale well to a fixed number of
CPUs? Will it scale well proportional to the number of subjects,
thetas, ...? Any hints on how many CPU/node would be optimal for a
cluster? Any hints on how well NONMEM will work with virtualized
multi-CPU machines?
Thanks,
Bill
Notice: This e-mail message, together with any attachments, contains
information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station,
New Jersey, USA 08889), and/or its affiliates Direct contact information
for affiliates is available at
http://www.merck.com/contact/contacts.html) that may be confidential,
proprietary copyrighted and/or legally privileged. It is intended solely
for the use of the individual or entity named on this message. If you are
not the intended recipient, and have received this message in error,
please notify us immediately by reply e-mail and then delete it from
your system.