Yeah, that will be a problem. We made large changes to the BTL interface with v2.x. If your customer requires a version that works with v1.10 I suggest looking that the usnic btl to see how it might be possible to write a btl that supports both the new and the old interface. You can also use an older template and then later update it for BTL v3. I have attached an older version of the template btl. It has the correct interfaces but may be somewhat out of date (we updated it for BTL v3).
You will not be able to use atomics with RMA in v1.10.x. It would take significant work to get a BTL v3 interface into v1.10 and port osc/rdma to v1.10. I don't know about oshmem. -Nathan HPC-5, LANL On Fri, Mar 04, 2016 at 12:37:47PM -0500, dpchoudh . wrote: > Hello Nathan, Mike and all > > Thank you for your responses. Let me rephrase them to make sure I > understood them correctly, and please correct me if I didn't: > > 1. Atomics are (have been) used in OSHMEM in the current (v1) release > 2. They are (will be) used in the MPI RMA in v2 release, which has not > happened yet > > I am sorry if I sound like I am nitpicking, but the reason I ask is that I > am trying to implement a new BTL that I am supposed to demo on a > customer's existing OMPI code base (which is obviously v1) but I am doing > the development out of the master branch (to avoid porting later), so I am > in a bit of spaghetti situation. > > Thank you > Durga > Life is complex. It has real and imaginary parts. > On Fri, Mar 4, 2016 at 11:06 AM, Nathan Hjelm <hje...@lanl.gov> wrote: > > On Thu, Mar 03, 2016 at 05:26:45PM -0500, dpchoudh . wrote: > > Hello all > > > > Here is a 101 level question: > > > > OpenMPI supports many transports, out of the box, and can be > extended to > > support those which it does not. Some of these transports, such as > > infiniband, provide hardware atomic operations on remote memory, > whereas > > others, such as iWARP, do not. > > > > My question is: how (and where in the code base) does openMPI use > this > > feature, on those hardware that support it? What is the penalty, in > terms > > of additional code, runtime performance and all other > considerations, on a > > hardware that does not support it? > > Network atomics are used for oshmem (see Mike's email) and MPI RMA. For > RMA they are exposed through the BTL 3.0 interface on the v2.x branch > and master. So far we have only really implemented compare-and-swap, > atomic add, and atomic fetch-and-add. Compare-and-swap and fetch-and-add > are required by our optimized RMA component (ompi/mca/osc/rdma). > -Nathan > > _______________________________________________ > devel mailing list > de...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/devel > Link to this post: > http://www.open-mpi.org/community/lists/devel/2016/03/18688.php > _______________________________________________ > devel mailing list > de...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/devel > Link to this post: > http://www.open-mpi.org/community/lists/devel/2016/03/18689.php
btl_template.tgz
Description: application/tar-gz
pgp3O02ULvhOI.pgp
Description: PGP signature
