Hello David Thank you for the reply.
-- We ran two benchmarks: " bfsBellmanFord.x10 " and " bfsDijkstra.x10 " present in IMSuite benchmark suite inside -- IMSuite/IMSuite_timing/IMSuite_Recursive/X10-FAC/bfsBellmanFord.x10 (same path for bfsDijkstra.x10) . -- Input used for testing can be found at -- IMSuite/IMSuite_Input/IMSuiteInput_64/inputbfsBellman_64_-spmax.txt (we tested using 64 and 128 inputs for bfsBellmanFord and 512 input bfsDijkstra) -- We used a 16 core machine and X10_NTHREADS set to 16. -- X10 version used x10-2.3.0 (as benchmarks are currently distributed in this version), Java -- jdk1.7.51, g++ versions tested -- (4.4.2 and 4.3) Compilation -- x10c++ bfsBellmanFord.x10 Running -- ./a.out -in inputbfsBellman_64_-spmax.txt Note -- We found frequency of segmentation fault to increase as input size was increased (say from 64 to 128) Regards Suyash From: "David P Grove" <gro...@us.ibm.com> To: "Mailing list for users of the X10 programming language" <x10-users@lists.sourceforge.net> Sent: Wednesday, August 13, 2014 7:45:26 PM Subject: Re: [X10-users] Regarding Native X10 suy...@cse.iitm.ac.in wrote on 08/13/2014 02:41:55 AM: > > We were trying to run the IMSuite Benchmarks with Native (C++) > backend. On running few of the recursive benchamrks such as > (bfsBellmanFord.x10, bfsDijkstra.x10) in some runs we came across > "segmentation fault". When we thought of executing these benchmarks > on Managed (Java) backend, they ran to completion and produced the > correct output. We tried testing the kernels several time on both > backends and found Managed X10 runtime more stable. > > Also we observed segmentation fault was more evident in kernels > utilizing "clocks" (in Native X10). We are unable to interpret this > issue and why it occurs only with Native X10 runtime. > Setup of our testing: IMSuite Recursive Kernels -- X10-FAC set -- > bfsBellmanFord.x10 , Input -- inputbfsBellmanFord_64_-spmax.txt > Hi, Could you send me a link to the code you are running and detailed compile/run instructions to try to duplicate the failure? I can try to take a look next week. thanks, --dave ------------------------------------------------------------------------------ _______________________________________________ X10-users mailing list X10-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/x10-users
------------------------------------------------------------------------------
_______________________________________________ X10-users mailing list X10-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/x10-users
