Sebastian, If possible, I strongly suggest you look into CUDA 4.1 r2 and using Rolf vandeVaart's MPI CUDA RDMA 3). Your life will be MUCH easier.
Having used GPUDirect1 in the last half of 2010, I can say it is a pain for the 9 - 14% gain in efficiency we saw. Ken On Fri, 2012-01-20 at 18:20 +0100, Sebastian Rinke wrote: > With > > > * MLNX OFED stack tailored for GPUDirect > * RHEL + kernel patch > * MVAPICH2 > > > it is possible to monitor GPUDirect v1 activities by means of > observing changes to values in > > > * /sys/module/ib_core/parameters/gpu_direct_pages > * /sys/module/ib_core/parameters/gpu_direct_shares > > > By setting CUDA_NIC_INTEROP=1 there are no changes anymore. > > > Is there a different way now to monitor if GPUDirect actually works? > > > Sebastian. > > > > On Jan 18, 2012, at 5:06 PM, Kenneth Lloyd wrote: > > > > > It is documented > > in > > http://developer.download.nvidia.com/compute/cuda/4_0/docs/GPUDirect_Technology_Overview.pdf > > set CUDA_NIC_INTEROP=1 > > > > > > From: devel-boun...@open-mpi.org [mailto:devel-boun...@open-mpi.org] On > > Behalf Of Sebastian Rinke > > Sent: Wednesday, January 18, 2012 8:15 AM > > To: Open MPI Developers > > Subject: Re: [OMPI devel] GPUDirect v1 issues > > > > Setting the environment variable fixed the problem for Open MPI with > > CUDA 4.0. Thanks! > > > > However, I'm wondering why this is not documented in the NVIDIA > > GPUDirect package. > > > > Sebastian. > > > > On Jan 18, 2012, at 1:28 AM, Rolf vandeVaart wrote: > > > > > > > > Yes, the step outlined in your second bullet is no longer > > necessary. > > > > Rolf > > > > > > From: devel-boun...@open-mpi.org [mailto:devel-boun...@open-mpi.org] On > > Behalf Of Sebastian Rinke > > Sent: Tuesday, January 17, 2012 5:22 PM > > To: Open MPI Developers > > Subject: Re: [OMPI devel] GPUDirect v1 issues > > > > Thank you very much. I will try setting the environment variable and > > if required also use the 4.1 RC2 version. > > > > To clarify things a little bit for me, to set up my machine with > > GPUDirect v1 I did the following: > > > > * Install RHEL 5.4 > > * Use the kernel with GPUDirect support > > * Use the MLNX OFED stack with GPUDirect support > > * Install the CUDA developer driver > > > > Does using CUDA >= 4.0 make one of the above steps redundant? > > > > I.e., RHEL or different kernel or MLNX OFED stack with GPUDirect > > support is not needed any more? > > > > Sebastian. > > > > Rolf vandeVaart wrote: > > > > I ran your test case against Open MPI 1.4.2 and CUDA 4.1 RC2 and it worked > > fine. I do not have a machine right now where I can load CUDA 4.0 drivers. > > Any chance you can try CUDA 4.1 RC2? There were some improvements in the > > support (you do not need to set an environment variable for one) > > http://developer.nvidia.com/cuda-toolkit-41 > > > > There is also a chance that setting the environment variable as outlined in > > this link may help you. > > http://forums.nvidia.com/index.php?showtopic=200629 > > > > However, I cannot explain why MVAPICH would work and Open MPI would not. > > > > Rolf > > > > > > > > -----Original Message----- > > From: devel-boun...@open-mpi.org [mailto:devel-boun...@open-mpi.org] > > On Behalf Of Sebastian Rinke > > Sent: Tuesday, January 17, 2012 12:08 PM > > To: Open MPI Developers > > Subject: Re: [OMPI devel] GPUDirect v1 issues > > > > I use CUDA 4.0 with MVAPICH2 1.5.1p1 and Open MPI 1.4.2. > > > > Attached you find a little test case which is based on the > > GPUDirect v1 test > > case (mpi_pinned.c). > > In that program the sender splits a message into chunks and sends > > them > > separately to the receiver which posts the corresponding recvs. It > > is a kind of > > pipelining. > > > > In mpi_pinned.c:141 the offsets into the recv buffer are set. > > For the correct offsets, i.e. increasing them, it blocks with Open > > MPI. > > > > Using line 142 instead (offset = 0) works. > > > > The tarball attached contains a Makefile where you will have to > > adjust > > > > * CUDA_INC_DIR > > * CUDA_LIB_DIR > > > > Sebastian > > > > On Jan 17, 2012, at 4:16 PM, Kenneth A. Lloyd wrote: > > > > > > > > Also, which version of MVAPICH2 did you use? > > > > I've been pouring over Rolf's OpenMPI CUDA RDMA 3 (using > > CUDA 4.1 r2) > > vis MVAPICH-GPU on a small 3 node cluster. These are > > wickedly interesting. > > > > Ken > > -----Original Message----- > > From: devel-boun...@open-mpi.org [mailto:devel-bounces@open- > > > > > > mpi.org] > > > > > > On Behalf Of Rolf vandeVaart > > Sent: Tuesday, January 17, 2012 7:54 AM > > To: Open MPI Developers > > Subject: Re: [OMPI devel] GPUDirect v1 issues > > > > I am not aware of any issues. Can you send me a test > > program and I > > can try it out? > > Which version of CUDA are you using? > > > > Rolf > > > > > > > > -----Original Message----- > > From: devel-boun...@open-mpi.org > > [mailto:devel-bounces@open- > > > > > > mpi.org] > > > > > > On Behalf Of Sebastian Rinke > > Sent: Tuesday, January 17, 2012 8:50 AM > > To: Open MPI Developers > > Subject: [OMPI devel] GPUDirect v1 issues > > > > Dear all, > > > > I'm using GPUDirect v1 with Open MPI 1.4.3 and > > experience blocking > > MPI_SEND/RECV to block forever. > > > > For two subsequent MPI_RECV, it hangs if the recv > > buffer pointer of > > the second recv points to somewhere, i.e. not at > > the beginning, in > > the recv buffer (previously allocated with > > cudaMallocHost()). > > > > I tried the same with MVAPICH2 and did not see the > > problem. > > > > Does anybody know about issues with GPUDirect v1 > > using Open MPI? > > > > Thanks for your help, > > Sebastian > > _______________________________________________ > > devel mailing list > > de...@open-mpi.org > > http://www.open-mpi.org/mailman/listinfo.cgi/devel > > > > > > ----------------------------------------------------------------------------------- > > This email message is for the sole use of the intended recipient(s) and may > > contain > > confidential information. Any unauthorized review, use, disclosure or > > distribution > > is prohibited. If you are not the intended recipient, please contact the > > sender by > > reply email and destroy all copies of the original message. > > ----------------------------------------------------------------------------------- > > > > _______________________________________________ > > devel mailing list > > de...@open-mpi.org > > http://www.open-mpi.org/mailman/listinfo.cgi/devel > > > > > > > > _______________________________________________ > > devel mailing list > > de...@open-mpi.org > > http://www.open-mpi.org/mailman/listinfo.cgi/devel > > > > > > _______________________________________________ > > devel mailing list > > de...@open-mpi.org > > http://www.open-mpi.org/mailman/listinfo.cgi/devel > > > > _______________________________________________ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel ============== Kenneth A. Lloyd, Jr. CEO - Director of Systems Science Watt Systems Technologies Inc. Albuquerque, NM US This e-mail is covered by the Electronic Communications Privacy Act, 18 U.S.C. 2510-2521 and is intended only for the addressee named above. It may contain privileged or confidential information. If you are not the addressee you must not copy, distribute, disclose or use any of the information in it. If you have received it in error please delete it and immediately notify the sender.
