Re: [OMPI users] Help with HPL and MPI
Thanks for your reply Brock, I appreciate the help. I have rebuilt the makefile with your suggestions and tried some other things but to no avail. One thing that I realized but not sure if this makes a difference is that I have installed an openmpi package and an mpich package so could that cause problems using the mpicc compiler and linking the different packages/libraries? I have rebuilt my make file and something interesting that I have noticed is that no matter what I change I still get the same error. I am always getting the following error: Makefile:47: Make.inc: No such file or directory *** No rule to make target 'Make.inc'. Stop I use the following command in my top directory, which is /hpl: make arch=Beowulf while my makefile is Make.Beowulf. I am still at a loss at can't figure out what to do next so I really appreciate the help, even though this may seem trivial. The make file and error log follow: Make.Beowulf SHELL = /bin/sh # CD = cd CP = cp LN_S = ln -s MKDIR = mkdir RM = /bin/rm -f TOUCH = touch # ARCH = Linux_x86_64 # - HPL Directory Structure / HPL library -- TOPdir = $(HOME)/hpl INCdir = $(TOPdir)/include BINdir = $(TOPdir)/bin/$(ARCH) LIBdir = $(TOPdir)/lib/$(ARCH) # HPLlib = $(LIBdir)/libhpl.a # - Message Passing library (MPI) -- MPdir = /usr MPinc =-I$(MPdir)/include/openmpi MPlib = $(MPdir)/lib64/openmpi/libmpi.so # - Linear Algebra library (BLAS or VSIPL) - LAdir = /usr/lib64 LAinc =-I$(MPdir)/include LAlib = $(LAdir)/libblas.so.3 $(LAdir)/atlas/libblas.so.3 # F2CDEFS = # HPL_INCLUDES = -I$(INCdir) -I$(INCdir)/$(ARCH) $(LAinc) $(MPinc) HPL_LIBS = $(HPLlib) $(LAlib) $(MPlib) # HPL_OPTS = -DHPL_CALL_CBLAS # -- HPL_DEFS = $(F2CDEFS) $(HPL_OPTS) $(HPL_INCLUDES) # - Compilers / linkers - Optimization flags --- CC = mpicc CCNOOPT = CCFLAGS = $(HPL_DEFS) -pipe -O3 -funroll-loops # LINKER = mpicc LINKFLAGS = $(CCFLAGS) # ARCHIVER = ar ARFLAGS = r RANLIB = echo Error File ( Condensed to show the last portion with the error code ) make -f Make.top refresh_tst arch=Beowulf make[1]: Entering directory `/hpl' cp makes/Make.matgen testing/matgen/Beowulf/Makefile cp makes/Make.timer testing/timer/Beowulf/Makefile cp makes/Make.pmatgen testing/pmatgen/Beowulf/Makefile cp makes/Make.ptimer testing/ptimer/Beowulf/Makefile cp makes/Make.ptest testing/ptest/Beowulf/Makefile make[1]: Leaving directory `/hpl' make -f Make.top build_src arch=Beowulf make[1]: Entering directory `/hpl' ( cd src/auxil/Beowulf; make ) make[2]: Entering directory `/hpl/src/auxil/Beowulf' Makefile:47: Make.inc: No such file or directory make[2]: *** No rule to make target `Make.inc'. Stop. make[2]: Leaving directory `/hpl/src/auxil/Beowulf' make[1]: *** [build_src] Error 2 make[1]: Leaving directory `/hpl' make: *** [build] Error 2
Re: [OMPI users] Help with HPL and MPI
Im taking this off list as its off topic. yes having both mpich and openmpi installed can cause problem with names of libs being the same at times. Try the following make clean arch=Beowulf And try again, Also make sure that your mpicc is the openmpi one. Brock Palen Center for Advanced Computing bro...@umich.edu (734)936-1985 On Apr 12, 2007, at 12:08 AM, sn...@comcast.net wrote: Thanks for your reply Brock, I appreciate the help. I have rebuilt the makefile with your suggestions and tried some other things but to no avail. One thing that I realized but not sure if this makes a difference is that I have installed an openmpi package and an mpich package so could that cause problems using the mpicc compiler and linking the different packages/libraries? I have rebuilt my make file and something interesting that I have noticed is that no matter what I change I still get the same error. I am always getting the following error: Makefile:47: Make.inc: No such file or directory *** No rule to make target 'Make.inc'. Stop I use the following command in my top directory, which is /hpl: make arch=Beowulf while my makefile is Make.Beowulf. I am still at a loss at can't figure out what to do next so I really appreciate the help, even though this may seem trivial. The make file and error log follow: Make.Beowulf SHELL = /bin/sh # CD = cd CP = cp LN_S = ln -s MKDIR = mkdir RM = /bin/rm -f TOUCH = touch # ARCH = Linux_x86_64 # - HPL Directory Structure / HPL library -- TOPdir = $(HOME)/hpl INCdir = $(TOPdir)/include BINdir = $(TOPdir)/bin/$(ARCH) LIBdir = $(TOPdir)/lib/$(ARCH) # HPLlib = $(LIBdir)/libhpl.a # - Message Passing library (MPI) -- MPdir = /usr MPinc =-I$(MPdir)/include/openmpi MPlib = $(MPdir)/lib64/openmpi/libmpi.so # - Linear Algebra library (BLAS or VSIPL) - LAdir = /usr/lib64 LAinc =-I$(MPdir)/include LAlib = $(LAdir)/libblas.so.3 $(LAdir)/atlas/libblas.so.3 # F2CDEFS = # HPL_INCLUDES = -I$(INCdir) -I$(INCdir)/$(ARCH) $(LAinc) $(MPinc) HPL_LIBS = $(HPLlib) $(LAlib) $(MPlib) # HPL_OPTS = -DHPL_CALL_CBLAS # -- HPL_DEFS = $(F2CDEFS) $(HPL_OPTS) $(HPL_INCLUDES) # - Compilers / linkers - Optimization flags --- CC = mpicc CCNOOPT = CCFLAGS = $(HPL_DEFS) -pipe -O3 -funroll-loops # LINKER = mpicc LINKFLAGS = $(CCFLAGS) # ARCHIVER = ar ARFLAGS = r RANLIB = echo Error File ( Condensed to show the last portion with the error code ) make -f Make.top refresh_tst arch=Beowulf make[1]: Entering directory `/hpl' cp makes/Make.matgen testing/matgen/Beowulf/Makefile cp makes/Make.timer testing/timer/Beowulf/Makefile cp makes/Make.pmatgen testing/pmatgen/Beowulf/Makefile cp makes/Make.ptimer testing/ptimer/Beowulf/Makefile cp makes/Make.ptest testing/ptest/Beowulf/Makefile make[1]: Leaving directory `/hpl' make -f Make.top build_src arch=Beowulf make[1]: Entering directory `/hpl' ( cd src/auxil/Beowulf; make ) make[2]: Entering directory `/hpl/src/auxil/Beowulf' Makefile:47: Make.inc: No such file or directory make[2]: *** No rule to make target `Make.inc'. Stop. make[2]: Leaving directory `/hpl/src/auxil/Beowulf' make[1]: *** [build_src] Error 2 make[1]: Leaving directory `/hpl' make: *** [build] Error 2 ___ users mailing list us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/users
[OMPI users] 1st IEEE/IFIP International Workshop on End-to-end Virtualization and Grid Management
(Apologies if you receive multiple copies of this email.) = EVGM 2007 1st IEEE/IFIP International Workshop on End-to-end Virtualization and Grid Management San José, California, USA - held as part of Manweek 2007 October 29 - November 2, 2007 = Virtualization allows layer n (upper layer) to function independently and effectively without being intrinsically tied to layer n-1 (lower layer). In OS virtualization, guest operating systems are isolated from physical computational resources. In server virtualization, application servers are isolated from physical server resources. In network virtualization, network services or applications are isolated from physical network resources. One of the major features of Grids is the sharing of heterogeneous resources, which can be facilitated by various levels of virtualization. In other words, Grids should provide holistic or end-to-end virtualization. It is obvious that proper resource management is key to providing effective virtualization. However, resource management for virtualization is a challenge (as opposed to typical resource management) in that there is no trivial way of virtualizing resources and presenting virtualized resources to the upper layers. It is even a bigger challenge when end-to-end (E2E) virtualization is considered, for example, when OS and server virtualization interplay with storage and network virtualization for providing an effective E2E virtualized environment. This workshop will offer a unique opportunity for researchers and practitioners to exchange ideas and experiences on problems, challenges, solutions and potential future research, as well as development issues, in this new field of E2E Virtualization and Grid Management. In addition to paper presentations, the workshop provides an intimate setting for discussion and debate through panels and group work. The authors are encouraged to submit original papers on topics related to the concepts described above, including, but not limited to: # OS resource management for virtualization (OS Virtualization) # Server resource management for virtualization (Server Virtualization) # Application resource management for virtualization (Application Virtualization) # Storage resource management for virtualization (Storage Virtualization) # Network resource management for virtualization (Network Virtualization) # Network Management resource management for virtualization (NM Virtualization) # Heterogeneous Grid VO resource management for virtualization (Grid Virtualization) # Interplay of various virtualization layers # Virtual resource scheduling # Virtual resource discovery # Virtual resource monitoring # Virtualization in clusters and high-performance computing environments # Virtualization for ERP, CRM, Healthcare, Financial applications SUBMISSION EVGM 2007 is using the JEMS Conference Management System to manage paper submission process. In order to submit, you must first create an JEMS account. JEMS system will e-mail you your password. You need to login to JEMS using the password you have received by email. JEMS will provide you with submission homepage where you can register your paper submission and make appropriate changes. Authors are requested to submit either long or short papers, strictly in LNCS format: - Long papers (up to 10 single-spaced single-column pages) - Short papers describing work-in-progress (up to 4 pages) Submissions exceeding the above mentioned page limits will not be reviewed. Papers must be submitted electronically through the JEMS conference management system at https://jems.sbc.org.br/evgm2007 (only PDF files are permitted). IMPORTANT DEADLINES # Submission: May 13th 2007 # Notification: July 6th 2007 # Camera ready: August 2nd 2007 WEBSITE: http://magellan.tssg.org/2007/evgm/cfp.php Email: manweek07...@manweek.org
Re: [OMPI users] openmpi and Torque
Jeff Squyres wrote: On Apr 11, 2007, at 8:08 AM, Bas van der Vlies wrote: The OMPI_CHECK_PACKAGE macro is a rather nasty macro that tries to reduce the replication of checking for a header then a library, then setting CFLAGS, LDFLAGS, LIBS, and all that. There are two components that use the TM libraries, so we have a centralized macro that sets the configuration data for the two components. The -I, -L, -ls are set in OMPI_CHECK_PACKAGE. In the OMPI_CHECK_TM macro, you probably want to check for pbs- config, then if that fails, run the OMPI_CHECK_PACKAGE stuff. Of course, you might want to run the OMPI_CHECK_PACKAGE code with CFLAGS and LDFLAGS set to what you get from pbs-config, to make sure the libraries actually work. The way we set flags is a bit weird -- you need to set the environment variables: $1_CPPFLAGS $1_CFLAGS $1_LDFLAGS $1_LIBS I have made a crude version that uses 'pbs-config' and check uses the OMPI_CHECK_PACKAGE macro but in the OMPI_CHECK_PACKAGE_HEADER macro /include is appended if it not /usr or /usr/include. So there must option to prevent this (pbs-config give the right value). Then i do not have to set the CPPFLAGS variable. OMPI_CHECK_PACKAGE_LIB: For the library name there is also -l prepended so i can not use the name form 'pbs-config --libs' and it as the same problem as with checking the header if it is not standard then /lib is appended. What about assigning $1_ directly and then passing flags that will be ignored to OMPI_CHECK_PACKAGE (e.g., "/usr")? I have not tried this at all -- just hacked this in emacs. So it may not even compile -- you'll need to test it. See new attachment. :-) --Jeff Squyres Cisco Systems Jeff, Maybe is was not clear enough ;-). The $1_ do not work i had to use CPPFLAGS instead of $1_CPPFLAGS else the test fails. On our site the torque library is found in /usr/lib so that can be found else we also must set the LDFLAGS flag. I have the right values of include/library directory but they are handled as directory prefix by the OMPI_CHECK_PACKAGE macro and not as directory. What i tried to accomplish was that the library name was set by a variable. So if torque decided to use another name for the library we do not have to change the macro again. Regards ___ users mailing list us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/users -- * * * Bas van der Vlies e-mail: b...@sara.nl * * SARA - Academic Computing Servicesphone: +31 20 592 8012 * * Kruislaan 415 fax:+31 20 6683167* * 1098 SJ Amsterdam * * * ompi_check_tm.m4 Description: application/m4
Re: [OMPI users] openmpi and Torque
On Apr 12, 2007, at 3:45 AM, Bas van der Vlies wrote:
Jeff Squyres wrote:
On Apr 11, 2007, at 8:08 AM, Bas van der Vlies wrote:
The OMPI_CHECK_PACKAGE macro is a rather nasty macro that tries
to reduce the replication of checking for a header then a
library, then setting CFLAGS, LDFLAGS, LIBS, and all that.
There are two components that use the TM libraries, so we have
a centralized macro that sets the configuration data for the
two components. The -I, -L, -ls are set in OMPI_CHECK_PACKAGE.
In the OMPI_CHECK_TM macro, you probably want to check for pbs-
config, then if that fails, run the OMPI_CHECK_PACKAGE stuff.
Of course, you might want to run the OMPI_CHECK_PACKAGE code
with CFLAGS and LDFLAGS set to what you get from pbs-config, to
make sure the libraries actually work. The way we set flags is
a bit weird -- you need to set the environment variables:
$1_CPPFLAGS
$1_CFLAGS
$1_LDFLAGS
$1_LIBS
I have made a crude version that uses 'pbs-config' and check uses
the OMPI_CHECK_PACKAGE macro but in the OMPI_CHECK_PACKAGE_HEADER
macro /include is appended if it not /usr or /usr/include. So
there must option to prevent this (pbs-config give the right
value). Then i do not have to set the CPPFLAGS variable.
OMPI_CHECK_PACKAGE_LIB:
For the library name there is also -l prepended so i can not use
the name form 'pbs-config --libs' and it as the same problem as
with checking the header if it is not standard then /lib is
appended.
What about assigning $1_ directly and then passing flags that
will be ignored to OMPI_CHECK_PACKAGE (e.g., "/usr")? I have not
tried this at all -- just hacked this in emacs. So it may not
even compile -- you'll need to test it. See new attachment. :-)
--Jeff Squyres
Cisco Systems
Jeff,
Maybe is was not clear enough ;-). The $1_ do not work i
had to use CPPFLAGS instead of $1_CPPFLAGS else the test fails. On
our site the torque library is found in /usr/lib so that can be
found else we also
must set the LDFLAGS flag.
I have the right values of include/library directory but they are
handled as directory prefix by the OMPI_CHECK_PACKAGE macro and not
as directory.
What i tried to accomplish was that the library name was set by a
variable. So if torque decided to use another name for the library
we do not have to change the macro again.
Sorry not to respond yesterday -- I was frantically writing a couple
of papers that desperately needed finishing. Anyway, yeah, I forgot
that OMPI_CHECK_PACKAGE did all that stuff (I don't know why I forgot
-- that's the whole point of OMPI_CHECK_PACKAGE). Anyway, I'd just
figure out what the CFLAGS, LDFLAGS, and LIBS required for TM are
(using pbs-config, of course), then call an AC_CHECK_HEADER and
AC_CHECK_LIB to make sure the headers and libs are ok. Then set $1_
{CFLAGS, LDFLAGS, LIBS} as needed and return. Just remember that if
you set something in CFLAGS or friends during the macro, that you
should remove it before the macro returns so that all the other
configure tests don't have those values.
Brian
[OMPI users] mpi tcp error...
What was the fix last time [openmpi 1.2 is in action below] ? [arnoldg@honest3 mpi]$ !mpirun mpirun --mca btl self,sm,tcp -np 4 -machinefile hosts allall_openmpi_icc 50 50 1000 [honest1][0,1,0][btl_tcp_endpoint.c:572:mca_btl_tcp_endpoint_complete_connect] connect() failed with errno=113 mpirun: killing job... mpirun noticed that job rank 0 with PID 21119 on node honest1 exited on signal 15 (Terminated). 3 additional processes aborted (not shown) Troy and I talked about this off-list and resolved that the issue was with the TCP setup on the nodes. But it is worth noting that we had previously fixed a bug in the TCP setup in 1.0.2 with respect to the SEGVs that Troy was seeing -- hence, when he tested the 1.0.3 prerelease tarballs, there were no SEGVs. -Galen + Galen Arnold, consulting group--system engineer National Center for Supercomputing Applications 1205 W. Clark St.(217) 244-3473 Urbana, IL 61801 arno...@ncsa.uiuc.edu
Re: [OMPI users] openmpi and Torque
Brian Barrett wrote:
On Apr 12, 2007, at 3:45 AM, Bas van der Vlies wrote:
Jeff Squyres wrote:
On Apr 11, 2007, at 8:08 AM, Bas van der Vlies wrote:
The OMPI_CHECK_PACKAGE macro is a rather nasty macro that tries
to reduce the replication of checking for a header then a
library, then setting CFLAGS, LDFLAGS, LIBS, and all that.
There are two components that use the TM libraries, so we have
a centralized macro that sets the configuration data for the
two components. The -I, -L, -ls are set in OMPI_CHECK_PACKAGE.
In the OMPI_CHECK_TM macro, you probably want to check for pbs-
config, then if that fails, run the OMPI_CHECK_PACKAGE stuff.
Of course, you might want to run the OMPI_CHECK_PACKAGE code
with CFLAGS and LDFLAGS set to what you get from pbs-config, to
make sure the libraries actually work. The way we set flags is
a bit weird -- you need to set the environment variables:
$1_CPPFLAGS
$1_CFLAGS
$1_LDFLAGS
$1_LIBS
I have made a crude version that uses 'pbs-config' and check uses
the OMPI_CHECK_PACKAGE macro but in the OMPI_CHECK_PACKAGE_HEADER
macro /include is appended if it not /usr or /usr/include. So
there must option to prevent this (pbs-config give the right
value). Then i do not have to set the CPPFLAGS variable.
OMPI_CHECK_PACKAGE_LIB:
For the library name there is also -l prepended so i can not use
the name form 'pbs-config --libs' and it as the same problem as
with checking the header if it is not standard then /lib is
appended.
What about assigning $1_ directly and then passing flags that
will be ignored to OMPI_CHECK_PACKAGE (e.g., "/usr")? I have not
tried this at all -- just hacked this in emacs. So it may not
even compile -- you'll need to test it. See new attachment. :-)
--Jeff Squyres
Cisco Systems
Jeff,
Maybe is was not clear enough ;-). The $1_ do not work i
had to use CPPFLAGS instead of $1_CPPFLAGS else the test fails. On
our site the torque library is found in /usr/lib so that can be
found else we also
must set the LDFLAGS flag.
I have the right values of include/library directory but they are
handled as directory prefix by the OMPI_CHECK_PACKAGE macro and not
as directory.
What i tried to accomplish was that the library name was set by a
variable. So if torque decided to use another name for the library
we do not have to change the macro again.
Sorry not to respond yesterday -- I was frantically writing a couple
of papers that desperately needed finishing. Anyway, yeah, I forgot
that OMPI_CHECK_PACKAGE did all that stuff (I don't know why I forgot
-- that's the whole point of OMPI_CHECK_PACKAGE). Anyway, I'd just
figure out what the CFLAGS, LDFLAGS, and LIBS required for TM are
(using pbs-config, of course), then call an AC_CHECK_HEADER and
AC_CHECK_LIB to make sure the headers and libs are ok. Then set $1_
{CFLAGS, LDFLAGS, LIBS} as needed and return. Just remember that if
you set something in CFLAGS or friends during the macro, that you
should remove it before the macro returns so that all the other
configure tests don't have those values.
Oke thanks for the explanation.
--
* *
* Bas van der Vlies e-mail: b...@sara.nl *
* SARA - Academic Computing Servicesphone: +31 20 592 8012 *
* Kruislaan 415 fax:+31 20 6683167*
* 1098 SJ Amsterdam *
* *
Re: [OMPI users] mpi tcp error...
Nevermind, found the answer in: http://www.open-mpi.org/community/lists/users/2005/10/0250.php "-mca btl_tcp_exclude eth1 " -Galen + Galen Arnold, consulting group--system engineer National Center for Supercomputing Applications 1205 W. Clark St.(217) 244-3473 Urbana, IL 61801 arno...@ncsa.uiuc.edu On Thu, 12 Apr 2007, Galen Arnold wrote: What was the fix last time [openmpi 1.2 is in action below] ? [arnoldg@honest3 mpi]$ !mpirun mpirun --mca btl self,sm,tcp -np 4 -machinefile hosts allall_openmpi_icc 50 50 1000 [honest1][0,1,0][btl_tcp_endpoint.c:572:mca_btl_tcp_endpoint_complete_connect] connect() failed with errno=113 mpirun: killing job... mpirun noticed that job rank 0 with PID 21119 on node honest1 exited on signal 15 (Terminated). 3 additional processes aborted (not shown) Troy and I talked about this off-list and resolved that the issue was with the TCP setup on the nodes. But it is worth noting that we had previously fixed a bug in the TCP setup in 1.0.2 with respect to the SEGVs that Troy was seeing -- hence, when he tested the 1.0.3 prerelease tarballs, there were no SEGVs. -Galen + Galen Arnold, consulting group--system engineer National Center for Supercomputing Applications 1205 W. Clark St.(217) 244-3473 Urbana, IL 61801 arno...@ncsa.uiuc.edu ___ users mailing list us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/users
Re: [OMPI users] OpenMPI for Windows?
Hi Brian, Thanks for the info. It was exactly what I needed to know. --- Rohit S. Singh Phone: (905) 366-6164 Software Developer Fax: (905) 273-9789 Logitech International S. A.Toll Free: 1-866-291-1505 Remote Control Business Unithttp://www.logitech.com/harmony Brian Barrett To v> cc Sent by: Subject users-bounces@ope Re: [OMPI users] OpenMPI for n-mpi.org Windows? 04/10/2007 11:52 AM Please respond to Open MPI Users On Mar 30, 2007, at 4:23 PM, rohit_si...@logitech.com wrote: > I'm somewhat new to OpenMPI, but I'm currently evaluating it as a > communications mechanism between Windows and Unix servers. > > I noticed that under your FAQs ( > http://www.open-mpi.org/faq/?category=supported-systems), it says: > > There are plans to support Microsoft Windows in the not-distant > future. > > When is the not-distant future? Is it in scope for this year? > Will the > Windows support require a Unix emulation layer like Cygwin? > > I do apologize if the information I'm requesting here is > confidential in > nature. The Windows support is planned, but we don't have a definite time table. Much of the work has been done, although there are still pieces missing. I believe remote launching is one of those pieces. The Windows support will be native and will not require Cygwin (which is part of what is taking us so long -- it's a bit project to move a UNIX-based code like Open MPI to Windows without sacrificing too much in maintainability and performance. Hope this helps, Brian ___ users mailing list us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/users
