Distribution Name	GCC Runtime
RedHat Enterprise Linux Server	4.4.6

History of problems fixed

The following fixes apply to IBM PE_RTE updates for both Linux x86 and Power Systems, unless noted otherwise.

IBM PE_RTE 1.3.0.11 [October 1, 2015]

MPI:

None.

MPICH:

None.

PAMI:

Critical fixes for Silent Data Corruption
General fixes

PDB:

None.

POE:

General fixes

IBM PE_RTE 1.3.0.10 not released for Power Platform

IBM PE_RTE 1.3.0.9 [October 30, 2014]

MPI:

None.

MPICH:

PE 1.3 - MEMORY LEAK ISSUES

PAMI:

Critical fixes for Silent data corruption in certain PE RTE PAMI and MPICH apps
AS OF 1.3.0.5 POE TASKS USE MORE VIRTUAL MEMORY
PE MP_BULK_XFER PROBLEM WITH OVER 64K TASKS

PDB:

None.

POE:

General fixes

IBM PE_RTE 1.3.0.8 [June 9, 2014]

MPI:

None.

MPICH:

MPI_WIN_POST AND MPI_WIN_COMPLETE COREDUMP or HANG

PAMI:

Critical fixes for Silent data corruption in some PAMI, MPICH, & OpenSHMEM apps and in PAMI fence operation
C BINDINGS IN FORTRAN INTEL COMPILER FAILS REGI
General fixes

PDB:

None.

POE:

MAKING OPENMP ABLE TO USE ALL ALLOCATED CPUS
mpicc enforcely uses GNU compiler
CODE COMPILED WITH MP_COMPILER=PGI FAILS AT RUN TIME.
General fixes

IBM PE_RTE 1.3.0.7 [December 9, 2013]

MPI:

None.

MPICH:

None.

PAMI:

Add CentOS support
General fixes

PDB:

General fixes

POE:

New scripts to generate F90 MPI module
General fixes

IBM PE_RTE 1.3.0.6 [September 13, 2013]

MPI:

Fixed problem during MPI job termination
Fixed a memory leak issue in MPI_Barrier when run in debug mode.

MPICH:

Critical fix: MPI_Reduce_scatter problem

TASK FLOODING CAUSES POOR MPI_REDUCE PERFORMANCE
Fixed problem using FORTRAN with MPI_Alltoall and MPI_IN_PLACE
Fixed problem when FCA is enabled
General fixes for collective selection
General fixes for dynamic tasking
General fixes for performance
Fix to prevent job failure if insufficient tokens

PAMI:

Critical fix: All to AllV algorithm data integrity

Fixed IP-job data integrity issue
Enable FCA on Power Linux
Fixed IB RDMA initialization
Fixed initialization problem with shared memory jobs
Fixed multi-endpoint issue during initialization
Fixed multiple sub-job issue
Fixed possible hang during job termination
General fixes for collective selection
Fixed possible job failure when running multiple protocols
Increased max number of shared memory tasks per node to 512
Fixed problem when running shared memory only
Fixed write of default PNSD configuration file
Fixed checkpoint issue when running shared memory only
Improved performance starting Infiniband jobs
Fixed problem with small local buffer size
General fixes for FCA
Fixed Openshmem memory registration issue

PDB:

Added pdb stat command options.
Fixed a problem debugging tasks using launch mode.

POE:

Fixed an issue with POE pulse timeouts when using MPMD programming model.
Fixed a problem with excessive GOMP_CPU_AFFINITY warning messages.
Added POE standalone affinity enhancements.
Display task count in PE-LSF job data debug output
Fixed an issue during the end of dynamic task jobs.
Added task count to PE-LSF debug output.
Fixed an internal issue with host based authentication for support of multiple releases.
Added adapter affinity support for single network.
Fixed an issue capturing affinity error messages.
Fixed an issue with PE compiler scripts writing to STDOUT instead of STDERR.
Fixed a problem starting 32-bit MPI applications from shell scripts.
Fixed an issue attempting to write a PMD log to full directory.

IBM PE_RTE 1.3.0.4 [June 14, 2013]

MPI:

None.

POE:

Fixed a problem with jobs failing when producing a large amount of output.

PAMI:

None.

PDB:

None.

IBM PE_RTE 1.3.0.3 [May 28, 2013]

MPI:

Possible data corruption with Linux RDMA applications that change malloc behavior.

POE:

None.

PAMI:

Possible data corruption with Linux RDMA applications that change malloc behavior.

PDB:

None.

IBM PE_RTE 1.3.0.2 [May 20, 2013]

MPI:

Fix an issue with MPI_File_write_at_all.
Fix collective selection issues.
Fixed MPI IO issues.
Fixed a data corruption issue for collective call with small messages.
Fixed issues with dynamic tasking.
Updated MP_POLLING_INTERVAL default value.
Fixed dynamic library binary compatibility issues.
Added support for asynchronous progress engine for timer/receive interrupts.
Fixed MPICH2 one-sided communication call.

POE:

Fixed several checkpoint/restart issues.
Updated PMD to dynamically open shared libraries where appropriate.
Modified MP_RFIFO_SIZE value range checks.
Fixed an issue during the end of dynamic task jobs.
Added collective selection support fixes.
Fixed a problem with missing line numbers in Linux lightweight core files.
Updated packaging & installation for support of multiple releases & rolling upgrades.

PAMI:

Fix an issue freeing window resource error handling.
Port deadlock detention.
Fixed a deadlock problem with multiple endpoints.
Added performance improvements for polling interval handling.
Fixed an issue with alltoallv memory violations.
Fixed an issue where 32-bit applications start slow when shared memory is used.
Fixed a hang condition in PAMI geometry.
Fix problem in M2MComposite alltoall algorithm.
Update example codes.
Fix PAMI_Fence issue when MP_RELIABLE_HW=yes.
Fix problem when multiple protocols are used.
Lift the FCA barrier restriction.
Fixed Alltoall(v) scaling issue with PGAS.
Corrected syscall numbers for RH 6.3.

PDB:

None.

IBM PE_RTE 1.3.0.1 [March 15, 2013]

MPI:

Rollup of fixes for PE RTE 1.3.0.0 GA.

POE:

Rollup of fixes for PE RTE 1.3.0.0 GA.

PAMI:

Rollup of fixes for PE RTE 1.3.0.0 GA.

PDB:

Rollup of fixes for PE RTE 1.3.0.0 GA.

Product/Component Name:	Platform:	Fix:
Parallel Environment Runtime Edition	Linux 64-bit,pSeries RHEL 6	PERTE-1.3.0.11-power-Linux

Readme and Release notes for release 1.3.0.11 Parallel Environment Runtime Edition 1.3.0.11 PERTE-1.3.0.11-power-Linux Readme