17/Jul/07: precompiled for Oct-2.9.12 and both LAM-7.1.3 and Open-MPI-1.2.3   mpitb-beta-FC6-OCT2912-LAM713-OMPI123.tar.bz2 (5062419bytes) (sum file oct2912lam713ompi123sum) PDFs comparing performance in Pi and PingPong subdirs... It seems a good idea to gradually move towards OMPI, not only for performance but also for maintainability.
	27/Jun/07: precompiled for both Oct-2.9.12 and Oct-2.1.73, with LAM-7.1.3  mpitb-beta-FC6-OCT2912-2173-LAM713.tar.bz2 (3883291 bytes) (sum file oct2912_2173lam713sum) PDFs comparing performance in Pi and PingPong subdirs... very small performance loss in 2.9.12. Confirmed correct 64-bit operation with ParallelKnoppix (see here).
	21/Jun/07: MPITB source code & precompiled version for an IA-32, FC6, Oct-2.9.10, LAM-7.1.3, system.  mpitb-beta-FC6-OCT2910-LAM713.tar.bz2 (2539552 bytes) (sum file oct2910lam713sum)  Save the sum file to the same subdir and type md5sum -c oct2910lam713sum to check for file integrity.   New demos, see the README file. No tutorial file anymore.  Recommending to use mpirun now, after seeing Yun Guan (alt) efforts to solve these cluster configuration problems. Thanks for the nice remark before section 2 (credit is on Michael's side)
	06/Jul/06: MPITB source code & precompiled versions for a couple of systems: for an Intel Xeon i686 with gcc-3.4.3, RedHat EL ES 4, Octave 2.1.73, LAM 7.1.2 for an Intel Pentium D x86-64 with gcc-4.1.0, Fedora Core 5, Octave 2.1.73, LAM 7.1.2 Adjust the DLD link to the precompiled version you want to use (points to the 32-bit version by default) mpitb-RHEL4FC5-OCT2173-LAM712.tgz (4915570 bytes) (sum file oct2173lam712sum) Preliminary versions of the NPB EP benchmark and the Mandelbrot example. Should work (after recompiling) on IA-64architectures as well. Thanks to Gianvito Quarta for providing access to his 64-node 128-cpu ItaniumII (biprocessor) cluster to test the toolbox. Itaniums are IA-64, a different architecture (EPIC) with different machine language (of the VLIW kind). We have no Itaniums, Gianvito let us use his cluster to adapt MPITB to IA-64. 64-bit Opterons and Pentiums are not IA-64, they are 64bit extensions of the IA-32 architecture, same machine language with longer registers / type sizes.
	11/May/05: MPITB source and precompiled code for Fedora Core 3, Octave 2.1.69, LAM 7.1.1 mpitb-FC3-OCT2169-LAM711.tgz (2507237 bytes) (sum file oct2169lam711sum) ( README and TUTORIAL files). Reworked Pi, Speedup, Spawn, Wavelets demos. Utilities subdirectory, including instrumentation and protocol aids.
	29/Oct/04: MPITB source and precompiled code for RedHat 8.0, Octave 2.1.60, LAM 7.1.1 mpitb-RH80-OCT2160-LAM711.tgz (1886697 bytes) (sum file oct2160lam711sum) ( README and TUTORIAL files). support for new integer types and ND arrays. Pi & Speedup demos with plotting OK. Spawn demo functional. Tutorial extended to cover packing of the new types, packing _on_ the new types, reduce operations with them.
	22/Apr/04: MPITB source and precompiled code for RedHat 8.0, Octave 2.1.57, LAM 7.0.4 mpitb-RH80-OCT2157-LAM704.tgz (1894805 bytes) (sum file oct2157lam704sum). plotting (Pi demo) broken
	20/Apr/04: MPITB source and precompiled code for RedHat 8.0, Octave 2.1.50, LAM 7.0.2 mpitb-RH80-OCT2150-LAM702.tgz (1305621 bytes). Please, save the oct2150lam702sum file to the same subdir and type md5sum -c oct2150lam702sum to check for file integrity. includes source code, Pi & Speedup demos, extensive tutorial file (see readme and tutorial)