Marco, Good point...
GLB (I know) and SatX10 (I guess) are written in a particular style that: - (i) enables the programs to run with a single thread - (ii) takes advantage of the single-thread context to remove all kinds of synchronization. While it is currently possible to do so, this is not something we recommend doing or commit to support in the future. Essentially, (ii) does not improve performance much and (i) is really hard to get right. I think we should refactor SatX10 and GLB to not depend on X10_STATIC_THREADS. But I don't know that we'll have the bandwidth to do this anytime soon. Olivier Marco Bungart <m.bung...@gmx.net> wrote on 05/23/2014 10:29:24 PM: > From: Marco Bungart <m.bung...@gmx.net> > To: x10-users@lists.sourceforge.net > Date: 05/23/2014 10:30 PM > Subject: Re: [X10-users] Resilient X10, async, X10_STATIC_THREADS > and DeadPlaceException > > Thanks for the answer. > Does that means, that programs like SatX10 or GLB (which suggest/need > X10_STATIC_THREADS=true) do not run correctly, if this variable is not > set and therefore are not supported? > > Thanks in advance, > > Marco > > > > Hi, > > > > The X10_STATIC_THREADS environment variable is not supported and, in > > general, will cause programs to freeze. > > The X10 runtime requires dynamic thread creation for all but the most > > trivial programs. > > > > Olivier > > > > Marco Bungart <m.bungart@...> wrote on 05/23/2014 07:56:30 AM: > > > > > From: Marco Bungart <m.bungart@...> > > > To: x10-users@... > > > Date: 05/23/2014 07:56 AM > > > Subject: [X10-users] Resilient X10, async, X10_STATIC_THREADS and > > > DeadPlaceException > > > > > Hi there, > > > > > I got non-deterministic behaviour with the code appended. Compiling > > > the > > > program with the c++ backend and starting it with 2 as parameter > > > results > > > almost always in the expected output: > > > > > > $ ./DPEExample 2 > > > Place(0): Sending to Place(1). > > > Place(0): waiting. > > > Place(0): Place(1) has answered. Ending. > > > Place(1): received message from Place(0). > > > Place(1): Sending to Place(2). > > > Place(1): waiting. > > > Place(1): Place(2) has answered. Ending. > > > Place(1): resetting Place(0). > > > Place(2): received message from Place(1). > > > Place(2): Sending to Place(3). > > > Place(2): waiting. > > > Place(2): Place(3) has answered. Ending. > > > Place(2): resetting Place(1). > > > Place(3): received message from Place(2). > > > Place(3): resetting Place(2). > > > Place 1 exited unexpectedly with exit code: 1 > > > Place(0): caught multiple exceptions, size = 2 > > > Place(0): Place(1) died. Resetting Place(0). > > > Place(0): Place(1) died. Resetting Place(0). > > > END. > > > $ > > > > > > Starting the same program with 3 as parameter, however, freezes the > > > program most of the time: > > > > > > $ ./DPEExample 3 > > > Place(0): Sending to Place(1). > > > Place(0): waiting. > > > Place(0): Place(1) has answered. Ending. > > > Place(1): received message from Place(0). > > > Place(1): Sending to Place(2). > > > Place(1): waiting. > > > Place(1): Place(2) has answered. Ending. > > > Place(1): resetting Place(0). > > > Place(2): received message from Place(1). > > > Place(2): Sending to Place(3). > > > Place(2): waiting. > > > Place(2): Place(3) has answered. Ending. > > > Place(2): resetting Place(1). > > > Place(3): received message from Place(2). > > > Place(3): resetting Place(2). > > > > > > [program freezes here] > > > > > > Compiling the program with the java backend changes the behaviour: > > > - Starting with 2 as parameter: ~50% of all runs show the "correct" > > > (first) behaviour, rest freezes > > > - Starting with 3 as parameter resulted for me always in a freeze > > > (second behaviour) > > > > > > Environment-variables used: > > > export X10_NPLACES=4 > > > export X10_NTHREADS=1 > > > export X10_STATIC_THREADS=true > > > export X10_RESILIENT_MODE=1 > > > export X10_RESILENT_NODE_ZERO=1 > > > > > > gcc used: 4.8.2 > > > java used: openJDK 1.7.0_55 > > > > > > Unsetting X10_STATIC_THREADS seems to fix the problem. > > > Am I doing something fundamentally wrong or is this a bug? Might it > > > be > > > related to Bug #3368 (http://jira.codehaus.org/browse/XTENLANG- > > > 3368)? > > > > > > Cheers and thanks in advance, > > > > > > Marco > > > [attachment "DPEExample.x10" deleted by Olivier Tardieu/Watson/IBM] > > > > > >------------------------------------------------------------------------------ > > > > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For > > > FREE > > > Instantly run your Selenium tests across 300+ browser/OS combos. > > > Get unparalleled scalability from the best Selenium testing > > > platformavailable > > > Simple to use. Nothing to install. Get started now for free." > > > > > http://p.sf.net > > /sfu/SauceLabs_______________________________________________ > > > > > X10-users mailing list > > > X10-users@... > > > https://lists.sourceforge.net/lists/listinfo/x10-users > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS combos. > Get unparalleled scalability from the best Selenium testing platformavailable > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > X10-users mailing list > X10-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/x10-users > ------------------------------------------------------------------------------ The best possible search technologies are now affordable for all companies. Download your FREE open source Enterprise Search Engine today! Our experts will assist you in its installation for $59/mo, no commitment. Test it for FREE on our Cloud platform anytime! http://pubads.g.doubleclick.net/gampad/clk?id=145328191&iu=/4140/ostg.clktrk _______________________________________________ X10-users mailing list X10-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/x10-users
