October 2014 - petsc-dev - mailman3.cels.anl.gov

SF broken in master !
by Barry Smith 14 Oct '14

14 Oct '14

and yet it is not broke in next. I thought that was impossible! ******* Testing: testexamples_C ******* 44,47c44,47 < 0: 4001 2000 2002 3002 4002 < 0: 1001 3000 < 0: 2001 4000 < 0: 3001 1000 --- > 0: 4002 0 0 0 0 > 0: 3000 0 > 0: 4000 0 > 0: 1000 0 /home/balay/petsc.clone/src/vec/is/sf/examples/tutorials Possible problem with ex1_4_basic, diffs above ========================================= 49,52c49,52 < 0: 4100 2000 < 0: 1100 3000 1200 < 0: 2100 4000 1201 < 0: 3100 1000 1202 --- > 0: 4000 2000 > 0: 1000 3000 1000 > 0: 2000 4000 1000 > 0: 3000 1000 1000 /home/balay/petsc.clone/src/vec/is/sf/examples/tutorials Possible problem with ex1_5_basic, diffs above ========================================= 48c48 < [0] 0 <- (3,1) --- > [0] 0 <- (3,0) 51c51 < [1] 0 <- (0,1) --- > [1] 0 <- (0,0) 53c53 < [1] 2 <- (0,2) --- > [1] 2 <- (0,0) 55c55 < [2] 0 <- (1,1) --- > [2] 0 <- (1,0) 57c57 < [2] 2 <- (0,3) --- > [2] 2 <- (0,0) 59c59 < [3] 0 <- (2,1) --- > [3] 0 <- (2,0) 61c61 < [3] 2 <- (0,4) --- > [3] 2 <- (0,0) 66,75c66,70 < [0] Number of roots=2, leaves=5, remote ranks=3 < [0] 0 <- (3,1) < [0] 1 <- (1,0) < [0] 2 <- (1,2) < [0] 3 <- (2,2) < [0] 4 <- (3,2) < [1] Number of roots=3, leaves=2, remote ranks=2 < [1] 0 <- (0,1) < [1] 1 <- (2,0) < [2] Number of roots=3, leaves=2, remote ranks=2 --- > [0] Number of roots=2, leaves=1, remote ranks=1 > [0] 0 <- (2,2) > [1] Number of roots=3, leaves=1, remote ranks=1 > [1] 0 <- (2,0) > [2] Number of roots=3, leaves=1, remote ranks=1 77,80c72,73 < [2] 1 <- (3,0) < [3] Number of roots=3, leaves=2, remote ranks=2 < [3] 0 <- (2,1) < [3] 1 <- (0,0) --- > [3] Number of roots=3, leaves=1, remote ranks=1 > [3] 0 <- (0,0) /home/balay/petsc.clone/src/vec/is/sf/examples/tutorials Possible problem with ex1_7_basic, diffs above =========================================

1 0

Re: [petsc-dev] change default KSP max iterations
by Jed Brown 14 Oct '14

14 Oct '14

We have to overhaul the test harness anyway. One way or another, we need to call a runner that is responsible for everything (including output verification). Whatever we choose will be simpler than the current. On October 14, 2014 9:13:19 AM CDT, Dave May <dave.mayhem23(a)gmail.com> wrote: >Or alternatively, the options set used to define the test should >include a >sensible value of -ksp_max_it and include the option >-ksp_converged_reason. >If divergence occurred, then the string written to stdout indicating >the >reason for termination would provide sufficient information (even with >diff) to determine that the test failed. > >However, I can appreciate that this would require a massive overhaul of >all >the current test suite. > >Given the current test infrastructure used by PETSc, how could one >could >easily impose a time limit on each test without making a huge mess >inside >the makefile? Putting all the options and bash used to verify the tests >inside the makefiles is already pretty hairy. > > >On 14 October 2014 14:15, Jed Brown <jed(a)jedbrown.org> wrote: > >> This can't be fixed simply by reducing 10000. After all, 250^2 is >larger >> and that's just one more level of nesting. >> >> We could set time limits on tests and kill them if they don't >complete. >> >> On October 14, 2014 6:28:12 AM CDT, Barry Smith <bsmith(a)mcs.anl.gov> >> wrote: >> > >> >On Oct 13, 2014, at 10:30 PM, Jed Brown <jed(a)jedbrown.org> wrote: >> > >> >> Barry Smith <bsmith(a)mcs.anl.gov> writes: >> >> >> >>> Does anyone object if I change the default KSP max iterations >from >> >10,000 to 250? >> >>> >> >>> The rational is that 10,000 is absurd and if you don’t see >> >convergence by 250 it generally is pretty hopeless. Users, of >course, >> >can set a higher max it. >> >> >> >> This depends a bit on the problem. Iterative methods for >> >high-frequency >> >> problems often need a lot of iterations, for example. >> >> >> >> My concern is mostly for people that only every once in a while >use >> >more >> >> than 250 iterations, the solve fails (confusingly because it "used >to >> >> work"), and they have to start over. Also that just to >demonstrate >> >> simple scaling behavior with an example like KSP ex2, you'll need >to >> >> bump up the max iterations. >> >> >> >> Does the marginal value of saving seconds/minutes(?) >> > >> >8+ hours is how long one of the (pretty small) nightly test examples >> >that used inner outer iterations (when someone screwed up the code >so >> >the outer did not converge). Normally the example took < 10 >iterations >> >and ran very quickly. >> > >> > >> > >> > >> >> for a solver to >> >> reach 10k iterations (when you forgot to enable enough monitoring >to >> >> know what's happening) offset the confusion and alleged >misbehavior >> >for >> >> the people that want to keep iterating? >> >>

1 0

Re: [petsc-dev] change default KSP max iterations
by Dave May 14 Oct '14

14 Oct '14

Or alternatively, the options set used to define the test should include a sensible value of -ksp_max_it and include the option -ksp_converged_reason. If divergence occurred, then the string written to stdout indicating the reason for termination would provide sufficient information (even with diff) to determine that the test failed. However, I can appreciate that this would require a massive overhaul of all the current test suite. Given the current test infrastructure used by PETSc, how could one could easily impose a time limit on each test without making a huge mess inside the makefile? Putting all the options and bash used to verify the tests inside the makefiles is already pretty hairy. On 14 October 2014 14:15, Jed Brown <jed(a)jedbrown.org> wrote: > This can't be fixed simply by reducing 10000. After all, 250^2 is larger > and that's just one more level of nesting. > > We could set time limits on tests and kill them if they don't complete. > > On October 14, 2014 6:28:12 AM CDT, Barry Smith <bsmith(a)mcs.anl.gov> > wrote: > > > >On Oct 13, 2014, at 10:30 PM, Jed Brown <jed(a)jedbrown.org> wrote: > > > >> Barry Smith <bsmith(a)mcs.anl.gov> writes: > >> > >>> Does anyone object if I change the default KSP max iterations from > >10,000 to 250? > >>> > >>> The rational is that 10,000 is absurd and if you don’t see > >convergence by 250 it generally is pretty hopeless. Users, of course, > >can set a higher max it. > >> > >> This depends a bit on the problem. Iterative methods for > >high-frequency > >> problems often need a lot of iterations, for example. > >> > >> My concern is mostly for people that only every once in a while use > >more > >> than 250 iterations, the solve fails (confusingly because it "used to > >> work"), and they have to start over. Also that just to demonstrate > >> simple scaling behavior with an example like KSP ex2, you'll need to > >> bump up the max iterations. > >> > >> Does the marginal value of saving seconds/minutes(?) > > > >8+ hours is how long one of the (pretty small) nightly test examples > >that used inner outer iterations (when someone screwed up the code so > >the outer did not converge). Normally the example took < 10 iterations > >and ran very quickly. > > > > > > > > > >> for a solver to > >> reach 10k iterations (when you forgot to enable enough monitoring to > >> know what's happening) offset the confusion and alleged misbehavior > >for > >> the people that want to keep iterating? > >

1 0

Re: [petsc-dev] change default KSP max iterations
by Jed Brown 14 Oct '14

14 Oct '14

This can't be fixed simply by reducing 10000. After all, 250^2 is larger and that's just one more level of nesting. We could set time limits on tests and kill them if they don't complete. On October 14, 2014 6:28:12 AM CDT, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >On Oct 13, 2014, at 10:30 PM, Jed Brown <jed(a)jedbrown.org> wrote: > >> Barry Smith <bsmith(a)mcs.anl.gov> writes: >> >>> Does anyone object if I change the default KSP max iterations from >10,000 to 250? >>> >>> The rational is that 10,000 is absurd and if you don’t see >convergence by 250 it generally is pretty hopeless. Users, of course, >can set a higher max it. >> >> This depends a bit on the problem. Iterative methods for >high-frequency >> problems often need a lot of iterations, for example. >> >> My concern is mostly for people that only every once in a while use >more >> than 250 iterations, the solve fails (confusingly because it "used to >> work"), and they have to start over. Also that just to demonstrate >> simple scaling behavior with an example like KSP ex2, you'll need to >> bump up the max iterations. >> >> Does the marginal value of saving seconds/minutes(?) > >8+ hours is how long one of the (pretty small) nightly test examples >that used inner outer iterations (when someone screwed up the code so >the outer did not converge). Normally the example took < 10 iterations >and ran very quickly. > > > > >> for a solver to >> reach 10k iterations (when you forgot to enable enough monitoring to >> know what's happening) offset the confusion and alleged misbehavior >for >> the people that want to keep iterating?

1 0

Re: [petsc-dev] change default KSP max iterations
by Barry Smith 14 Oct '14

14 Oct '14

On Oct 13, 2014, at 10:30 PM, Jed Brown <jed(a)jedbrown.org> wrote: > Barry Smith <bsmith(a)mcs.anl.gov> writes: > >> Does anyone object if I change the default KSP max iterations from 10,000 to 250? >> >> The rational is that 10,000 is absurd and if you don’t see convergence by 250 it generally is pretty hopeless. Users, of course, can set a higher max it. > > This depends a bit on the problem. Iterative methods for high-frequency > problems often need a lot of iterations, for example. > > My concern is mostly for people that only every once in a while use more > than 250 iterations, the solve fails (confusingly because it "used to > work"), and they have to start over. Also that just to demonstrate > simple scaling behavior with an example like KSP ex2, you'll need to > bump up the max iterations. > > Does the marginal value of saving seconds/minutes(?) 8+ hours is how long one of the (pretty small) nightly test examples that used inner outer iterations (when someone screwed up the code so the outer did not converge). Normally the example took < 10 iterations and ran very quickly. > for a solver to > reach 10k iterations (when you forgot to enable enough monitoring to > know what's happening) offset the confusion and alleged misbehavior for > the people that want to keep iterating?

1 0

Re: [petsc-dev] change default KSP max iterations
by Richard Mills 14 Oct '14

14 Oct '14

I've seen many cases where 250 iterations is not necessarily an indicator of anything being wrong. But, yes, 10K is ridiculous. Why not set the default max to 1000? That's an order of magnitude lower than the current default but is unlikely (I think) to cause people to start seeing puzzling failures in codes that previously worked OK with the current default. --Richard On Oct 14, 2014 12:30 AM, "Lisandro Dalcin" <dalcinl(a)gmail.com> wrote: > On 14 October 2014 06:30, Jed Brown <jed(a)jedbrown.org> wrote: > > Barry Smith <bsmith(a)mcs.anl.gov> writes: > > > > Does the marginal value of saving seconds/minutes(?) for a solver to > > reach 10k iterations (when you forgot to enable enough monitoring to > > know what's happening) offset the confusion and alleged misbehavior for > > the people that want to keep iterating? > > Following your reasoning, is 10k enough? While 250 is probably too > small, I still think that 10k is too much. > > -- > Lisandro Dalcin > ============ > Research Scientist > Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) > Numerical Porous Media Center (NumPor) > King Abdullah University of Science and Technology (KAUST) > http://numpor.kaust.edu.sa/ > > 4700 King Abdullah University of Science and Technology > al-Khawarizmi Bldg (Bldg 1), Office # 4332 > Thuwal 23955-6900, Kingdom of Saudi Arabia > http://www.kaust.edu.sa > > Office Phone: +966 12 808-0459 >

1 0

Re: [petsc-dev] change default KSP max iterations
by Lisandro Dalcin 14 Oct '14

14 Oct '14

On 14 October 2014 06:30, Jed Brown <jed(a)jedbrown.org> wrote: > Barry Smith <bsmith(a)mcs.anl.gov> writes: > > Does the marginal value of saving seconds/minutes(?) for a solver to > reach 10k iterations (when you forgot to enable enough monitoring to > know what's happening) offset the confusion and alleged misbehavior for > the people that want to keep iterating? Following your reasoning, is 10k enough? While 250 is probably too small, I still think that 10k is too much. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Numerical Porous Media Center (NumPor) King Abdullah University of Science and Technology (KAUST) http://numpor.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 4332 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] change default KSP max iterations
by Jed Brown 13 Oct '14

13 Oct '14

Barry Smith <bsmith(a)mcs.anl.gov> writes: > Does anyone object if I change the default KSP max iterations from 10,000 to 250? > > The rational is that 10,000 is absurd and if you don’t see convergence by 250 it generally is pretty hopeless. Users, of course, can set a higher max it. This depends a bit on the problem. Iterative methods for high-frequency problems often need a lot of iterations, for example. My concern is mostly for people that only every once in a while use more than 250 iterations, the solve fails (confusingly because it "used to work"), and they have to start over. Also that just to demonstrate simple scaling behavior with an example like KSP ex2, you'll need to bump up the max iterations. Does the marginal value of saving seconds/minutes(?) for a solver to reach 10k iterations (when you forgot to enable enough monitoring to know what's happening) offset the confusion and alleged misbehavior for the people that want to keep iterating?

1 0

[PATCH] Fix buffer overrun in threaded MatZeroEntries_SeqBAIJ
by Lawrence Mitchell 13 Oct '14

13 Oct '14

rmap->trstarts counts rows in the matrix, but a->i only has nrows/bs entries. Fix start and end indexing by dividing by the block size. While we're here, update the sizeof call to use MatScalar rather than PetscScalar. --- I think I sent this to petsc-maint a while ago, but either it didn't make it, or slipped through the cracks. Anyway, here goes again. If configured with threading, MatZeroEntries on a BAIJ Mat indexes too far into a->i, the fix is just to divide appropriately by the block size, I think. I wonder if the sizeof change should instead be to: sizeof(*(a->a)) so that one doesn't have to update again if the Mat datatype changes in the future. src/mat/impls/baij/seq/baij2.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/mat/impls/baij/seq/baij2.c b/src/mat/impls/baij/seq/baij2.c index aa1ed44..46d1cac 100644 --- a/src/mat/impls/baij/seq/baij2.c +++ b/src/mat/impls/baij/seq/baij2.c @@ -2709,10 +2709,10 @@ PetscErrorCode MatZeroEntries_SeqBAIJ_Kernel(PetscInt thread_id,Mat A) PetscInt n,start,end; Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data; - start = trstarts[thread_id]; - end = trstarts[thread_id+1]; + start = trstarts[thread_id]/A->rmap->bs; + end = trstarts[thread_id+1]/A->rmap->bs; n = a->i[end] - a->i[start]; - ierr = PetscMemzero(a->a+a->bs2*a->i[start],a->bs2*n*sizeof(PetscScalar));CHKERRQ(ierr); + ierr = PetscMemzero(a->a+a->bs2*a->i[start],a->bs2*n*sizeof(MatScalar));CHKERRQ(ierr); return 0; } -- 2.1.0

1 0

Fwd: See how your site did in September with Google Analytics
by Matthew Knepley 13 Oct '14

13 Oct '14

---------- Forwarded message ---------- From: Google Analytics <analytics-noreply(a)google.com> Date: Mon, Oct 13, 2014 at 6:29 AM Subject: See how your site did in September with Google Analytics To: knepley(a)gmail.com [image: Google Analytics] <http://www.google.com/appserve/mkt/p/aFuEN6DIVIw8fAb35iiXmRvhzTqAuKJrVIbkPw…> Your Customer ID: petsc.org Sign in <http://www.google.com/appserve/mkt/p/OOU0Zt5ZZlLoV01eIJSo9m-40GLKlyIwW-MaER…> A snapshot of your website in September. Take a few minutes to review last month's website performance for *petsc.org*. We hope that these key metrics and our monthly tip will help you identify new opportunities for your business. VISITS 1,203 The total number of visits to your site this month BOUNCE RATE 45% Your bounce rate shows how many people leave your site without visiting any other pages on your site TIME ON SITE 04:35 The average amount of time someone spends on your site in a single session AVERAGE PAGES 3 The average number of pages viewed in a single session Did you know you had *1,203* <http://www.google.com/appserve/mkt/p/F_SO_btZ4vWRMGiWrGfsqQWg4hv--bk-1LQQfE…> visits this month? Keep an eye on your All Traffic report <http://www.google.com/appserve/mkt/p/vuU5g3GNUXKB0ayIesSAc6Zgb0ioYpADm3fFOn…> to see exactly where your visitors come from so you can make tweaks to your marketing campaigns. Sign in <http://www.google.com/appserve/mkt/p/l8SyFaN_U2ES7zy43owf_ze06ohXuWZ6BbK0EC…> Your Audience TOP COUNTRIES TOP COUNTRY *503 Visits* United States *214 Visits* Germany *91 Visits* France *74 Visits* United Kingdom *50 Visits* Spain Your Location report <http://www.google.com/appserve/mkt/p/qqxUapmLqB5tsqPR65DWs_DbuMvrV4_fhn6KNf…> tells you where your visitors live around the world. Make sure you’re promoting your business in all the right areas. View your Location report <http://www.google.com/appserve/mkt/p/q5TLbXbfcxgey4Sc5dqsh1sNNGG-0oBwWUiZUL…> UNIQUE VISITORS 315 New Visitors 116 Return Visitors 431 Unique Visitors [image: New Visitors / Return Visitors] How people are finding you TOP KEYWORDS bitbucket formatting 16 Visits https://bitbucket.org/pets ... 3 Visits https://bitbucket.org/pflo ... 1 Visits petsc4py bitbucket 1 Visits using the explicit runge-k ... 1 Visits ADWORDS VISITS 1 AdWords Visitors 1,202 Non-AdWords Visitors 1,203 Total Visitors [image: New Visitors / Return Visitors] View your AdWords account <http://www.google.com/appserve/mkt/p/DQS6uAe7b82cLJG9HIe-2s9RYckrQv7zgP1P1Z…> TOP SOCIAL SOURCES 83% reddit 17% Google+ 0% 0% 0% TOP REFERRERS google / organic *335 Visits* (direct) / (none) *297 Visits* mcs.anl.gov / referral *233 Visits* fenicsproject.org / referral *106 Visits* lists.mcs.anl.gov / referral *83 Visits* What’s happening when users come to your site TOP PAGES Visits /petsc/petsc/c ... /petsc/petsc /petsc/petsc/c ... /petsc/petsc/o ... /petsc/petsc/p ... Pageviews Page Exits 239 EXITS Did you know that *239 (19.9%)* people exited your site at /petsc/petsc/commits/all <http://www.google.com/appserve/mkt/p/KyO2mvnxYcCqek1tnT0VbQrLnNwQGfcAcos_ls…>? View your Exit Pages <http://www.google.com/appserve/mkt/p/V3qWe_n0VC6o9tKifn-464lL417yCpkgxsswsO…> Your Goals NO GOALS SET UP YET 0 Use goals to track customer interactions that matter to your business the most, such as purchases or number of pages viewed. ABANDONMENT RATE 0% The percentage of people who start but do not complete one of your goals Your goals completion sources Looks like you don’t have any Goals set up yet. Use Goals to keep an eye on customer actions that matter to you, like clicks and sales. Set up Goals <http://www.google.com/appserve/mkt/p/gnWiyBw3YPkaA4NmRg3ProjF4ScEtDmXb58e_V…> Thank you for reviewing our report. If you have questions please see our more detailed reporting overview <http://www.google.com/appserve/mkt/p/w-H9UfiIcn0ZMOiP8fZdv8fJlOpr76O8ANoAm-…>. And please take a few minutes to provide feedback <http://www.google.com/appserve/mkt/p/Y5tdWRij9a2GwxcPL4GkZVSh2zrXN7oyBRUD_s…>. Go to your account <http://www.google.com/appserve/mkt/p/SR34LOCq7lkQC2k4Uu1h_CqPj4T3Lwexl9J2Is…> <http://www.google.com/appserve/mkt/p/rTyZmqkV3YPHDtyPOGlYY60Kda0M2HzoC_4msD…> Join the Community <http://www.google.com/appserve/mkt/p/Fjuy1asMkLL9vq5o0O7kXAQeJ48Dap0RbLZmJQ…> <http://www.google.com/appserve/mkt/p/wcU4SnTstqMDBfgwtEeMZqiQ12SNI3vWXz_UaG…> Ask Experts <http://www.google.com/appserve/mkt/p/xfm_cc5PyldfA_a1K1A2-vMNjELcJPUD8pKH2C…> <http://www.google.com/appserve/mkt/p/UeH3qC5_VC3l_1n05OILq-1VFEc2lGT5bYCkKU…> Learn <http://www.google.com/appserve/mkt/p/OBU83sFHyRFx-kwtynChRt_W8-JAZQ3P2CRRmB…> © 2014 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043 This email was sent to knepley(a)gmail.com because you indicated that you'd like to receive updates and tips about your Google Analytics account. If you don't want to receive such emails in the future, please unsubscribe here <http://www.google.com/appserve/mkt/optout/0b5bL9f2r5hNgcAtUwfIDRRWCl1nIaiYX…>. You can also change your preferences on your account's Google Analytics Email Communications page by logging into https://www.google.com/analytics/settings/my_account <http://www.google.com/appserve/mkt/p/wNyHe6MP7WhX4WK9fzpIma1qNcD8sNBDHAZ_Mi…> -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0