[openib-general] Announcing the Release of MVAPICH2 0.9.3 with multi-threading support and anonymous SVN access

[Dhabaleswar Panda]

The MVAPICH team is pleased to announce the availability of MVAPICH2
0.9.3 with the following new features:

 - Multi-threading support: This support is available for Gen2, VAPI
   and uDAPL transport interfaces. In addition, multi-threading
   support for TCP/IP interface (provided by MPICH2 stack) is also
   available.

 - Integrated with MPICH2 1.0.3 stack

 - Advanced AVL tree-based Resource-aware registration cache

 - Tuning and Optimization of various collective algorithms for a wide
   range of system sizes

 - Processor affinity for intra-node shared memory communication

 - Auto-detection of InfiniBand adapters for Gen2 

MVAPICH2 0.9.3 release supports Gen2, VAPI and uDAPL transport
interfaces. It also has support for the standard TCP/IP (provided by
MPICH2 stack). It is optimized for the following platforms, OS,
compilers and InfiniBand adapters:

  - Platforms: EM64T, Opteron, IA-32, PPC and Mac G5 
  - Operating Systems: Linux, Solaris and Mac OSX 
  - Compilers: gcc, intel, pathscale and pgi 
  - InfiniBand Adapters: 
       - Mellanox adapters with PCI-X and PCI-Express
         (SDR and DDR with mem-full and mem-free cards)
       - PathScale adapter (through OpenIB/Gen2 support)
       - IBM ehca adapter (through OpenIB/Gen2 support)

More details on all features and supported platforms can be obtained
by visiting the project's web page -> Overview -> features.

Starting with this 0.9.3 release, MVAPICH team is also pleased to
announce the availability of the MVAPICH2 code base through anonymous
SVN access. Nightly tarballs are also available. The mvapich-commit
mailing list can also be used by users, developers and vendors to keep
track of all commits happening to the SVN.

MVAPICH2 0.9.3 continues to deliver excellent performance.  Sample
performance numbers include:
 
  - OpenIB/Gen2 on EM64T with PCI-Ex and IBA-DDR:
      Two-sided operations: 
        - 3.28 microsec one-way latency (4 bytes)
        - 1475 MB/sec unidirectional bandwidth 
        - 2661 MB/sec bidirectional bandwidth  

      One-sided operations:
        - 4.99 microsec Put latency 
        - 1476 MB/sec unidirectional Put bandwidth 
        - 2661 MB/sec bidirectional Put bandwidth 

  - OpenIB/Gen2 on EM64T with PCI-Ex and IBA-SDR:
      Two-sided operations:
        - 3.71 microsec one-way latency (4 bytes) 
        - 964 MB/sec unidirectional bandwidth
        - 1846 MB/sec bidirectional bandwidth 

      One-sided operations:
        - 6.12 microsec Put latency
        - 964 MB/sec unidirectional Put bandwidth
        - 1846 MB/sec bidirectional Put bandwidth

  - OpenIB/Gen2 on Opteron with PCI-Ex and IBA-SDR: 
      Two-sided operations:
        - 3.38 microsec one-way latency (4 bytes) 
        - 971 MB/sec unidirectional bandwidth
        - 1867 MB/sec bidirectional bandwidth 

      One-sided operations:
        - 5.98 microsec Put latency
        - 971 MB/sec unidirectional Put bandwidth
        - 1867 MB/sec bidirectional Put bandwidth

  - Solaris uDAPL/IBTL on Opteron with PCI-Ex and IBA-SDR:
      Two-sided operations:
        - 5.41 microsec one-way latency (4 bytes)
        - 981 MB/sec unidirectional bandwidth
        - 1903 MB/sec bidirectional bandwidth

      One-sided operations:
        - 7.42 microsec Put latency
        - 981 MB/sec unidirectional Put bandwidth
        - 1903 MB/sec bidirectional Put bandwidth

  - OpenIB/Gen2 uDAPL on Opteron with PCI-Ex and IBA-SDR:
      Two-sided operations:
        - 3.61 microsec one-way latency (4 bytes)
        - 971 MB/sec unidirectional bandwidth
        - 1894 MB/sec bidirectional bandwidth

      One-sided operations:
        - 6.10 microsec Put latency
        - 971 MB/sec unidirectional Put bandwidth
        - 1894 MB/sec bidirectional Put bandwidth

Performance numbers for all other platforms, system configurations and
operations can be viewed by visiting `Performance' section of the
project's web page.

Additional features of MVAPICH2 0.9.3 release include:

- Similar performance with MVAPICH: With the ADI-3-level design,
  MVAPICH2 0.9.3 delivers similar performance for two-sided operations
  compared to MVAPICH 0.9.7. Organizations and users interested in
  getting the best performance for both two-sided and one-sided 
  operations and also want to exploit `multi-threading' capability may
  migrate from MVAPICH code base to MVAPICH2 code base. 

- A set of benchmarks to evaluate both two-sided and one-sided
  operations (Put, Get, and Accumulate). A new micro-benchmark 
  (Multi-threaded Latency Test) has been added.

- An enhanced and detailed `User Guide' is now available (in both html
  and pdf forms) from the FAQ page.

For downloading MVAPICH2 0.9.3 package and accessing the anonymous
SVN, please visit the following URL:

http://nowlab.cse.ohio-state.edu/projects/mpi-iba/

A stripped down version of this release is also available at the
OpenIB SVN.

All feedbacks, including bug reports and hints for performance tuning,
are welcome. Please post it to the mvapich-discuss mailing list.

Thanks, 

MVAPICH Team at OSU/NBCL 

======================================================================
MVAPICH/MVAPICH2 project is currently supported with funding from
U.S. National Science Foundation, U.S. DOE Office of Science,
Mellanox, Intel, Cisco Systems, Sun Microsystems and Linux Networx;
and with equipment support from Advanced Clustering, AMD, Apple,
Appro, Dell, IBM, Intel, Mellanox, Microway, PathScale, SilverStorm
and Sun Microsystems. Other technology partner includes Etnus.
======================================================================