[OMPI devel] CUDA kernels in OpenMPI

I'm trying to build a CUDA kernel into OpenMPI (because I'm experimenting with an Allreduce collective with data in GPU buffers, and I want the GPU to do the reduction). This involves writing a '.cu' file, and compiling this to '.o' with the NVIDIA CUDA compiler 'nvcc'; and also writing some of

Re: [OMPI devel] CUDA kernels in OpenMPI

It's hard to tell without complete makefile example. Could you please post a minimal reprocase? Note specifically for OpenMPI there is a tricky workaround. You can use nvcc as mpicc compiler by exporting OMPI_CC=nvcc and wrapping out incompatible compiler options. Kind regards, - Dmitry Mikushin.

[OMPI devel] Reminder: assign as well as request review

Hey folks Just a reminder. If you request a review from someone, GitHub doesn’t show that person’s icon when looking at the list of PRs. It only shows their icon and marks the PR with their ID if you actually “assign” it to that person. Thus, just requesting a review without assigning the PR to

[OMPI devel] CUDA kernels in OpenMPI

Here is the complete Makefile so far. I have it in directory ompi/mca/coll/ibm , which contains an implementation of an IBM-written collectives library. It won't work as-is, because I don't know how to use 'libtool' which is presumably needed to do the compile. If anybody can show me a rule us

[OMPI devel] Problem on master

Hello all There is a known issue on master that we are attempting to debug. Sadly, it is one that only shows on multi-node operations, and the signature varies based on your environment. We hope to have this resolved soon (and no, it doesn’t appear to be due to any one specific commit). In the

[OMPI devel] CUDA kernels in OpenMPI

It looks like the mailing system deleted the attachment, so here it is inline # # Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana # University Research and Technology # Corporation. All rights reserved. # Copyright (c) 2004-20

Re: [OMPI devel] CUDA kernels in OpenMPI

Hi Chris, First, you will need to have some configure stuff to detect nvcc and use it inside your Makefile. UTK may have some examples to show here. For the C/C++ API, you need to add 'extern "C"' statements around the interfaces you want to export in C so that you can use them inside Open MP

Re: [OMPI devel] Reminder: assign as well as request review

Ralph, It looks like GitHub *might* have rolled out the solution to your problem just this week: https://github.com/blog/2306-filter-pull-request-reviews-and-review-requests This appears to include an "Awaiting review from you" filter. Not quite a dashboard or notification, but at least a way to

[OMPI devel] [2.0.2rc4] "make install" failure on NetBSD/i386 (libtool?)

I had no problem with 2.0.2rc3 on NetBSD, but with 2.0.2rc4 I am seeing a "make install" failure (below). This is seen on an x86 (32-bit) platform, but not x86_64. I cannot say for certain that this is an Open MPI regression, since there *have* been s/w updates on this system since I last tested.

Re: [OMPI devel] Reminder: assign as well as request review

Thanks Paul - that does indeed help! > On Jan 27, 2017, at 12:26 PM, Paul Hargrove wrote: > > Ralph, > > It looks like GitHub *might* have rolled out the solution to your problem > just this week: > > https://github.com/blog/2306-filter-pull-request-reviews-and-review-requests >

Re: [OMPI devel] Reminder: assign as well as request review

I am so often the guy complaining about what it busted. So, it feels nice to have contributed something *positive* on this list. -Paul On Fri, Jan 27, 2017 at 5:42 PM, r...@open-mpi.org wrote: > Thanks Paul - that does indeed help! > > On Jan 27, 2017, at 12:26 PM, Paul Hargrove wrote: > > Ral