Thanks for the comments. Responses below.
Terry D. Dontje wrote:
Do you mean the RSL API? Or do you mean the APIs of alternative runtime
I think the concept is a good idea. A few questions that come to mind:
1. Do you have a set of APIs you plan on supporting?
The rsl API is in
As far as other runtime systems, I have not looked too much at what
others support. However, I am trying to make the APIs in the RSL as
generic as possible.
Not in the sense of new functionality, but some of the APIs are quite
different then ORTE is currently using.
2. Are you planning on adding new APIs (not currently supported by ORTE)?
3. Do any of the ORTE replacement APIs differ in how they work?
Well, every runtime does things differently.
For instance, looking at the MPICH PMI interface (which is sort-of their
version of the RSL), they make heavy use of a key-value space. For the
RSL, I am using process attributes which are similar in concept to this,
but do work slightly differently.
Another difference is that the RSL exposes a out of band communication
interface, which is not provided by the PMI. So if we used a runtime
that was based on the PMI, then we would have to do our own out-of-band
communication within the RSL component.
Yes, although there is already a layer of abstraction here since the GPR
usage in OMPI all goes through the modex code.
4. Will RSL change in how we access information from the GPR? If not
how does this layer really separate us from ORTE?
So what would happen with the RSL would be that the modex send/recv
would be called, which would then call the process attribute send/recv
code. Alternatively, the process attribute system could be called directly.
The process attribute system in the RSL would then use whatever
implementation specific system it wants to exchange the data.
That is up to the rsl implementation. An out-of-band interface is
provided, and it is the components job to make sure the message is
5. How will RSL handle OOB functionality (routing of messages)?
Again, it is up to the implementation. OMPI assumes that all process
names it sees uniquely identify a remote process. In this sense, a
global process namespace would need be needed. But if the rsl wanted to
do some trickery to avoid the need for a global namespace, it probably
6. How does making the process names opaque differ from how ORTE
names processes? Do you still need a global namespace for a
I like the idea but I really wonder if this will even be half-baked in
1.3 (same concern as Jeff's).
Tim Prins wrote:
WHAT: Solicitation of feedback on the possibility of adding a runtime
services layer to Open MPI to abstract out the runtime.
WHY: To solidify the interface between OMPI and the runtime environment,
and to allow the use of different runtime systems, including different
versions of ORTE.
WHERE: Addition of a new framework to OMPI, and changes to many of the
files in OMPI to funnel all runtime request through this framework. Few
changes should be required in OPAL and ORTE.
WHEN: Development has started in tmp/rsl, but is still in its infancy. We hope
to have a working system in the next month.
I am working on creating an interface between OMPI and the runtime system.
This would make a RSL framework in OMPI which all runtime services would be
accessed from. Attached is a graphic depicting this.
This change would be invasive to the OMPI layer. Few (if any) changes
will be required of the ORTE and OPAL layers.
At this point I am soliciting feedback as to whether people are
supportive or not of this change both in general and for v1.3.
The current model used in Open MPI assumes that one runtime system is
the best for all environments. However, in many environments it may be
beneficial to have specialized runtime systems. With our current system this
is not easy to do.
With this in mind, the idea of creating a 'runtime services layer' was
hatched. This would take the form of a framework within OMPI, through which
all runtime functionality would be accessed. This would allow new or
different runtime systems to be used with Open MPI. Additionally, with such a
system it would be possible to have multiple versions of open rte coexisting,
which may facilitate development and testing. Finally, this would solidify the
interface between OMPI and the runtime system, as well as provide
documentation and side effects of each interface function.
However, such a change would be fairly invasive to the OMPI layer, and
needs a buy-in from everyone for it to be possible.
Here is a summary of the changes required for the RSL (at least how it is
1. Add a framework to ompi for the rsl, and a component to support orte.
2. Change ompi so that it uses the new interface. This involves:
a. Moving runtime specific code into the orte rsl component.
b. Changing the process names in ompi to an opaque object.
c. change all references to orte in ompi to be to the rsl.
3. Change the configuration code so that open-rte is only linked where needed.
Of course, all this would happen on a tmp branch.
The design of the rsl is not solidified. I have been playing in a tmp branch
(located at https://svn.open-mpi.org/svn/ompi/tmp/rsl) which everyone is
welcome to look at and comment on, but be advised that things here are
subject to change (I don't think it even compiles right now). There are
some fairly large open questions on this, including:
1. How to handle mpirun (that is, when a user types 'mpirun', do they
always get ORTE, or do they sometimes get a system specific runtime). Most
likely mpirun will always use ORTE, and alternative launching programs would
be used for other runtimes.
2. Whether there will be any performance implications. My guess is not,
but am not quite sure of this yet.
Again, I am interested in people's comments on whether they think adding
such abstraction is good or not, and whether it is reasonable to do such a
thing for v1.3.
devel mailing list
devel mailing list