Hi, Matyas - I'm afraid when we first developed DX in the late '80s to early '90s we didn't anticipate the move to 64 bit architectures and built a 2G limit into DX, which you are now hitting. A few years ago I built a version which uses 64 bit addresses, so arenas can grow past that limit, but since DX does a lot of address arithmetic in 32 bit integers, no single object (that is, Array) can be larger than 2G. While it would be possible to find and fix all the places that happens, it'd be very painful and error-prone.
If you'd like to try the version I 'fixed', its on SourceForge - google SourceForge OpenDX. It might work for you; I'd suggest you partition the data immediately after loading it. It also uses threading, rather than multiple processes, to take advantage of all your processor cores. Alternatively, there are some pretty good alternatives; personally, I use Paraview, but VisIt (google LLNL visit) has believers as well. Both will take advantage of distributed systems, spreading the computation across multiple nodes using MPI. Good luck! Greg From: opendx2-users-boun...@lists.berlios.de [mailto:opendx2-users-boun...@lists.berlios.de] On Behalf Of herein mátyás Sent: Monday, March 03, 2014 9:05 AM To: opendx2-users@lists.berlios.de Subject: [OpenDX2-users] dx_memory Dear dx Users, I have very large dx files (~800 MB; output from a mantle convection code) and I tried to use opendx to visualize the 3d temeprature field. But I could not since there is a memory error. The communication window says: out of memory reached limit in large arena. I have 32 GB of rams in my workstation and I also ran the dx with the command flag -memory 32000 but the large item in the arena never exceeds the 2 GB (1920 MB - exactly). I think there is a memory bug. Is there any other options to tell dx to use more memory than 2 GB ? Any suggestions are very welcome. This project is a part of my Phd. thanks; Matyas. -- Herein Matyas ELTE TTK Geofizikai es Urtudomanyi Tanszek Eotvos University Department of Geophysics and Space Sciences 1117 Budapest, Pázmány Péter sétány 1/C http://geophysics.elte.hu/
_______________________________________________ OpenDX2-users mailing list OpenDX2-users@lists.berlios.de https://lists.berlios.de/mailman/listinfo/opendx2-users
