Mark,
Can you use these additional configure options and try again.
--with-threadcomm --with-pthreadclasses
Shri
On Nov 8, 2013, at 11:24 AM, Mark Adams wrote:
I have not been able to get checkout PETSc on Mira with:
[adams@miralac1 ~]$ git clone g...@bitbucket.org:petsc/petsc.git
variable that we manage currently.
On a side note, an additional complexity to think about is if the user wants to
set different solver/user options for each thread. One possibility to handle
this is to use prefix -thread#num_option_name for it but it looks clunky.
Shri
For example in this case
this. The offsets are integers, and I think they should
be.
I was unclear. The offset value should be in bytes from the beginning of
the array, currently it is in ints from the beginning of the array. Ints is
just plan confusing, Shri has to have this strange size sizeof(his
struct)/sizeof
the beginning of the array. Ints
is just plan confusing, Shri has to have this strange size sizeof(his
struct)/sizeof(PetscInt) to indicate how big his chunk is.
No, its has nothing to do with Int. Where did this come from? In fact, I
use it all the time to mean Scalar offsets
On Mar 23, 2013, at 12:55 PM, Jed Brown wrote:
On Sat, Mar 23, 2013 at 11:40 AM, Shri abhyshr at mcs.anl.gov wrote:
I had added -pc_reuse specifically for use with fieldsplit since I wanted
only a part of the preconditioner to be built only once, for e.g.
-fieldsplit_0_pc_reuse
to be experimental
feature. If this feature is useful then I can add an API for it otherwise I can
take it out.
Shri
On Mar 23, 2013, at 11:25 AM, Jed Brown wrote:
On Fri, Mar 22, 2013 at 10:14 AM, Mark F. Adams mark.adams at columbia.edu
wrote:
I thought the primary mission of options_left
I don't need this feature immediately and don't think I can devote the time for
it over the next three months. I can work on this probably next year starting
April. I've added it to bitbucket issues.
Shri
On Dec 13, 2012, at 2:06 PM, Jed Brown wrote:
Okay, who is working on this? (Is it me
given time via TSPostStep() by calling
TSSetConvergedReason() with the flag TS_CONVERGED_USER.
Shri
- Original Message -
On Mon, Dec 10, 2012 at 8:36 PM, Barry Smith bsmith at mcs.anl.gov
wrote:
I'd the ability to provide to TS a list of times at which I'd like
the
ODE
into PetscFinalize that ensures that this stuff has been freed
(very messy if user does anything weird with communicators)
3. add way to mark PetscTrMalloc'd memory as indirect so that it is not
reported by default
We can go with 1 for now??
Shri
Will this work for threadcomm since it is not a PetscObject?
Shri
On Dec 8, 2012, at 11:56 AM, Barry Smith wrote:
I don't think any of these three things are reasonable approaches. You are
basically saying we should not log some memory usage because it is annoying.
Much better
Satish,
What's happening with the private repos hosted on petsc.cs.iit.edu? I had a
private repo there for my Ph.D. project (ts_emt). Although I don't need it
anymore, I just wanted to check if it was backed up somehow.
Thanks,
Shri
On Nov 18, 2012, at 9:43 PM, Satish Balay wrote:
On Sat
. What configure
options did you use?
Shri
On Oct 31, 2012, at 11:41 AM, Nystrom, William D wrote:
Shri,
Have you had a chance to investigate the issues related to the new PETSc
threads
package and MKL?
Dave
From: petsc-dev-bounces at mcs.anl.gov
There are no external packages and MKL is using GNU compilers
-lmkl_intel_lp64 -lmkl_gnu_thread -lmkl_core -lgomp -lpthread instead of Intel
compiler that you have -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -liomp5
-lpthread
Shri
On Oct 31, 2012, at 1:35 PM, John Fettig wrote:
On Wed, Oct 31, 2012
On Oct 26, 2012, at 10:58 AM, Nystrom, William D wrote:
Jed or Shri,
Are there other preconditioners I could use/try now with the petsc thread
support besides jacobi?
I looked around in the documentation for something like least squares
polynomial preconditioning
that is referenced
,
Karli
On 10/26/2012 10:58 AM, Nystrom, William D wrote:
Jed or Shri,
Are there other preconditioners I could use/try now with the petsc thread
support besides jacobi?
I looked around in the documentation for something like least squares
polynomial preconditioning
that is referenced
, now that I have some time,
and get back to you.
Thanks,
Shri
On Oct 25, 2012, at 8:05 PM, John Fettig wrote:
Hi Dave,
On Thu, Oct 25, 2012 at 5:21 PM, Nystrom, William D wdn at lanl.gov wrote:
Hi John,
I have also been trying to test out the pthreads and openmp support in
petsc-dev
PetscCommGetThreadComm explicitly NOT
collective. Any objections?
Shri
On Oct 20, 2012, at 5:02 PM, Jed Brown wrote:
On Sat, Oct 20, 2012 at 4:54 PM, Shri abhyshr at mcs.anl.gov wrote:
On Oct 20, 2012, at 3:58 PM, Jed Brown wrote:
Worse, validity of the cache changes whether it has a barrier (in debug
mode) or not.
I don't understand the point you
Barry, Jed,
Please see the attached patch based on Barry's suggestions. I tested this with
MPI and MPIUNI and did not see any memory leaks. Let me know what you think.
Thanks,
Shri
-- next part --
A non-text attachment was scrubbed...
Name: threadcomm.patch
Type
specific
implementations for each attribute.
PetscCommAttachAttribute(MPI_Comm comm,MPI_Keyval,void *attr)
PetscCommDetachAttribute(MPI_Comm,MPI_keyval)
One hiccup i see is for attributes that have their own reference counting, for
e.g., threadcomm and Hong's elemental grid stuff.
Thanks,
Shri
On Jul 17, 2012, at 10:06 PM, Barry Smith wrote:
On Jul 17, 2012, at 9:58 PM, Shri wrote:
On Jul 17, 2012, at 4:22 PM, Barry Smith wrote:
On Jul 17, 2012, at 2:24 PM, Shri wrote:
On Jul 17, 2012, at 1:17 PM, Barry Smith wrote:
Why all this messy attribute check
to zero then does the delete as usual.
This is exactly what is done, but PETSC_COMM_WORLD and PETSC_COMM_SELF
outlive PetscFinalize, so we remove one more reference count for each of
those so that it actually gets collected.
But Shri indicated there is a possible problem, what is that problem
On Jul 15, 2012, at 5:26 PM, Shri wrote:
On Jul 15, 2012, at 4:57 PM, Barry Smith wrote:
On Jul 15, 2012, at 4:48 PM, Shri wrote:
On Jul 15, 2012, at 3:27 PM, Barry Smith wrote:
Shri,
You need to test your code WITHOUT pthreads OR openmp configure.
That is what i am doing
options do i need to have a 32 bit build on x86_64?
CC=gcc -m32 FC=gfortran -m32?
Thanks,
Shri
also 32bit linux build gives errors with -fp_trap in ddot(). Haven't checked
why.
Satish
On Sat, 14 Jul 2012, Barry Smith wrote:
Many of them with make test
Running test examples
On Jul 15, 2012, at 3:27 PM, Barry Smith wrote:
Shri,
You need to test your code WITHOUT pthreads OR openmp configure.
That is what i am doing, no openmp or pthreads.
You have the call to PetscThreadCommFinalize() protected by PTHREADCLASSES
There is no PetscThreadCommFinalize
Which example are you running? I don't see any memory leak for the examples
i've tested (threadcomm examples sys/threadcomm/examples/tutorials/ and
snes/examples/tutorials/ex19.c,ex5.c)
On Jul 14, 2012, at 4:22 PM, Barry Smith wrote:
Please always test for memory leaks before pushing
[
http://www.youtube.com/watch?v=j1tbMW_Gxc4 ;))
for the scalable (parallel) solution of scientific applications, and
ideally suited for those modeled by partial differential equations. or
something of that sort.
Shri
. and Draganescu, A. and Hill, J. and
Ghattas, O. and van Bloemen Waanders, B.},
journal={Proceedings of the 2005 ACM/IEEE conference on Supercomputing},
year={2005},
publisher={IEEE Computer Society Washington, DC, USA}
}
On Wed, Jun 6, 2012 at 5:21 PM, Shri abhyshr at mcs.anl.gov wrote:
Hi
Guided Neurosurgery, Simon K. Warfield, Matthieu Ferrant, Xavier Gallez, Arya
Nabavi, Ferenc A. Jolesz, and Ron Kikinis, Proceedings of SC2000, 2000.
Please let me know if there are any more.
Thanks,
Shri
On May 14, 2012, at 4:08 PM, Jed Brown wrote:
On Sat, May 12, 2012 at 8:11 PM, Shri abhyshr at mcs.anl.gov wrote:
On Sat, May 12, 2012 at 6:40 AM, Shri abhyshr at mcs.anl.gov wrote:
1. I needed functions in sched.h like CPU_SET and others for setting the
core affinities and these are only
On Sat, May 12, 2012 at 6:40 AM, Shri abhyshr at mcs.anl.gov wrote:
1. I needed functions in sched.h like CPU_SET and others for setting the core
affinities and these are only active if __USE_GNU is set and i wasn't sure
whether _GNU_SOURCE or __USE_GNU should be defined directly, so i put
Stefano,
Barry added additional guard to ensure that the thread code is not active.
Please pull and retry. Sorry for the trouble, we are experimenting with the
thread code so some things might get broken in the process.
Shri
On May 6, 2012, at 6:24 AM, Stefano Zampini wrote:
I've just
I pushed some changes to fix these errors. Can you please pull and retry. -
Original Message -
I had pulled some hours ago from
http://petsc.cs.iit.edu/petsc/petsc-dev/rev/61898f35c333 and obtained
the following error on make test
[0]PETSC ERROR: PetscCommDuplicate() line 200 in
- Original Message -
What is left to be done before a release?
- Replace 2 DMMG examples
- Rework thread code
This should hopefully be done by friday.
Shri
Barry
I'm guessing its Sean. - Original Message -
Does anyone actually take credit for this anywhere on the site?
- Peter
On Tue, Mar 20, 2012 at 2:29 PM, Jed Brown jedbrown at mcs.anl.gov
wrote:
On Tue, Mar 20, 2012 at 14:27, Barry Smith bsmith at mcs.anl.gov
wrote:
Once we get
I am going.
- Original Message -
Please let me know if you are attending the SIAM PP.
Thanks
Barry
A non-text attachment was scrubbed...
Name: make.log
Type: text/x-log
Size: 41469 bytes
Desc: not available
URL:
http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110407/333ee286/attachment.bin
-fieldsplit_xxx_lag_preconditioner niterations which
will delay the refactorization for field xxx by niterations and if
niterations
is set to -1 then it will not update the factorization.
Thanks,
Shri
Barry
On Mar 14, 2011, at 4:41 PM, Shri wrote:
| A B | |dx| |fx
-fieldsplit_xxx_lag_preconditioner niterations which
will delay the refactorization for field xxx by niterations and if
niterations
is set to -1 then it will not update the factorization.
Thanks,
Shri
Barry
On Mar 14, 2011, at 4:41 PM, Shri wrote:
| A B | |dx
lag_preconditioner functionality be
moved to ksp or maybe a -fieldsplit_xxx_lag_preconditioner option?
Thanks,
Shri
- Original Message -
Shri,
In theory the flag for KSPSetOperators() (for example
SAME_PRECONDITIONER) is propagated down through the PCFIELDSPLIT into
the PC that is used on the A block. So in at least simple
circumstances if you use -snes_lag_preconditioner it should just
work
http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-current/docs/manualpages/Mat/MatStoreValues.html#MatStoreValues
Does this still exist? If so, is there any example?
Thanks,
Shri
? Anything in CUBLAS? I think
CUBLAS provides triangular solves, anything about LU factorization?
Thanks,
Shri
- Original Message -
No worries,
In that case, the presentation is a little out of date so I'll just mention a
few things
relevant files: aijcuda.cu/mpiaijcuda.cu
)
row 1: (1, 4)
row 2: (2, 4)
row 3: (3, 4)
row 4: (4, 4)
Shri
- Barry Smith bsmith at mcs.anl.gov wrote:
I have added PetscObjectPrintClassNamePrefixType() which is now (in theory)
used uniformly in all the XXXView ASCII info formats to display this
information appropriately
Boyce,
I've added the MPIAIJ version of MatAXPY to petsc-dev which does the
preallocation before doing the addition. Please test it with your code and
report any problems if you have.
Thanks,
Shri
- Shri abhyshr at mcs.anl.gov wrote:
- Boyce Griffith griffith at cims.nyu.edu
and it could be printed out in mat_view_info or
even better if in some way the variable name can be used directly.
What do you think?
Shri
] PetscHeaderCreate_Private() line 25 in src/sys/objects/inherit.c
[0] MatCreate() line 67 in src/mat/utils/gcreate.c
I suspect it might be due to MatHeaderReplace(). Any cues?
Boyce,can you please check if this working for your application.
Shri
- Boyce Griffith griffith at cims.nyu.edu
- Boyce Griffith griffith at cims.nyu.edu wrote:
On 9/27/10 5:28 PM, Shri wrote:
Barry,
I've modified MatAXPY_SeqAIJ as follows when
DIFFERENT_NONZERO_STRUCTURE flag is true (i) create a new matrix and do
preallocation for it. For determining the number of nonzeros per
2010/9/3 Jed Brown jed at 59a2.org
On Fri, 3 Sep 2010 09:15:09 -0600 (GMT-06:00), Shri abhyshr at mcs.anl.gov
wrote:
Leo,
Jed and I both tried running a MatLoad example (see
src/dm/da/examples/tests/ex35.c) with petsc configured in debug mode and
had issues
Pushed. Please let me know if you get any errors.
Shri
- Barry Smith wrote:
if (size 1) { /* change viewer to display matrix in natural ordering
*/ierr = MatShellSetOperation(A, MATOP_VIEW, (void (*)(void))
MatView_MPI_DA);CHKERRQ(ierr);/* turn off loading of matrix
ShriBarry
On Sep 2, 2010, at 10:17 AM, Leo van Kampenhout wrote:
Hello Shri,
thanks for the fix. The error is now gone but the loading of the Mat takes now
infinitely long.. help!!
Leo
2010
/problem-with-gprof-on-os-x-program-is-not-of-the-host-architecture
make ex35/Users/Shri/petsc-dev/osx-debug/bin/mpicxx -o ex35.o -c -Wall
-Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -pg
-I/Users/Shri/petsc-dev/osx-debug/include -I/Users/Shri/petsc-dev/include
-I/usr/X11/include -I
Thanks Jed.I've pushed ex35 to the repository (src/dm/da/examples/tests). You
can run it with the following options
mpiexec -n 2 ./ex35 -X 100 -Y 100 -Z 100 . Let me know if you can run the
example on a grid size of 202 X 102 X 102.
Shri
P.S : You might need to do make in src/dm/da/utils
entries from disk in
2 seconds is acceptable,what do you think?
Thanks,
Shri
Linux kunyang 2.6.35-ARCH #1 SMP PREEMPT Fri Aug 20 22:49:24 CEST 2010 x86_64
Intel(R) Core(TM)2 Duo CPU P8700 @ 2.53GHz GenuineIntel GNU/Linux
Jed
Shri - looks like merit should be PetscReal - not PetscComplex [pusing
a fix for this. If this is incorrect - you can revert my change]
That's correct...thanks.I am building petsc with complex right now to see if
there are any other changes needed.
Satish
I'm not sure whether DVI should have support for complex type since complex
variables with real upper and lower bounds don't make sense to me and i don't
know of any actual problems which have or need complex bounds.
- Barry Smith bsmith at mcs.anl.gov wrote:
Shri,
Since
Sorry about that...removed the header reading from VecLoadIntoVector_Binary by
mistake.
Thanks,
Shri
- Jed Brown jed at 59A2.org wrote:
On Sat, 5 Jun 2010 19:09:48 -0500, Barry Smith bsmith at mcs.anl.gov wrote:
Anybody monkeyed with DA or SDA basic functions?
dm/da/examples
I'm getting an error while compiling ksp examples.
bash-3.2$ make ex32
/Users/Shri/petsc-dev/osx-debug-real/bin/mpicxx -o ex32.o -c -Wall
-Wwrite-strings -Wno-strict-aliasing -g
-I/Users/Shri/petsc-dev/osx-debug-real/include -I/Users/Shri/petsc-dev/include
-I/usr/X11/include -I/Users/Shri
59 matches
Mail list logo