Hexagon has updated software/libraries. Users should recompile their applications to gain any stability or performance improvements.

MPI
xt-mpt 5.2.0 -> 5.2.1

Compilers, wrappers and debuggers
PGI 11.1.0 -> 11.2.0
xt-asyncpe 4.8 -> 4.9
ATP 1.1.1 -> 1.1.2

Libraries
Petsc 3.1.04 -> 3.1.05
xt-libsci 10.5.0 -> 10.5.01
TPSL 1.0.0

NOTES:

xt-mpt
Bugfixes.
767319, 768355 Fix adi reference counter problem for builtin datatypes
768385 Add better PMI debug for an ALPs get_rank_from_pipe failure
755921 Add PMI feature that gives SPMD support to DMAPP applications
767095 Add PMI feature to deliver rank and PID attribute file for debuggers

xt-libsci
Purpose:
--------
The LibSci 10.5.01 release adds new C++ interfaces for CRAFFT
providing users an option to use CRAFFT Serial and Distributed
Routines in C++ applications. These interfaces are provided via
C++ function overloading, allowing a user to call a simple, generic
CRAFFT routine name and use optional arguments for advanced
functionality if required. CRAFFT offers a simpler interface for
FFT routines to improve application developer productivity. In
some cases the performance of the CRAFFT distributed transforms
is 10-50% better than FFTW2 MPI transforms. Users requiring more
information on usage should see the intro_crafft manpage.

Also included in the LibSci 10.5.01 release are new generic
interfaces for C users of CASE. CASE is a collection of
simplified interfaces into LAPACK and ScaLAPACK style routines
introduced in LibSci 10.5.00 that solve the symmetric eigenproblem.
The generic interfaces are provided via C++ function overloading,
allowing a user to call a simple CASE routine name and use optional
arguments for advanced functionality if required. This simplifies
calling CASE from C, improving application developer productivity.
Refer to the 'intro_case' man page for more information.

Known Problems:
---------------
Bug 768141 slatms code seg faults with gnu and xtpe-mc12 loaded
When using CASE interfaces and compiling with gnu add
"-Wl,-whole-archive -lpthread -Wl,-no-whole-archive" to the
link line of your code. This fixes a known link issue with gcc.

Bug 767985 10.5.0 pgi libsci_quadcore_mp.so: undefined reference
to `pgf90_dealloc03'
The workaround is to use PGI 10.6.0 or higher

PETSc
petsc/3.1.05 module now automatically loads the tpsl module
which includes Sundials2.4.0, Hypre 2.6.0b, SuperLU 4.0,
SuperLU_DIST 2.3, MUMPS 4.9.2, Parmetis 3.1.1 and Scotch 5.1.
PETSc 3.1.05 is not supported if the TPSL module is unloaded.

The Cray PETSc 3.1.05 is equivalent to the official
patch release of PETSc-3.1-p5 by Argonne National Laboratory.

TPSL
TPSL(Third Party Scientific Libraries) contains a collection of
third party mathematical libraries for use with PETSc and Trilinos.
These libraries increase the flexibility of PETSc and Trilinos by
providing users with multiple options for solving problems in dense
and sparse linear algebra. The tpsl module is automatically loaded
when PETSc or Trilinos is loaded.

The libraries included are Hypre, SuperLU, SuperLU_dist, MUMPs,
ParMetis, Sundials, and Scotch.

* MUMPS 4.9.2. MUMPS (MUltifrontal Massively Parallel sparse
direct Solver) is a package of parallel, sparse, direct
linear-system solvers based on a multifrontal algorithm. For
further information, see http://graal.ens-lyon.fr/MUMPS/. MUMPS
can now interface with SCOTCH as well.

* SuperLU 4.0. SuperLU is a sequential version of SuperLU_dist
(not included with petsc-complex), and a sequential incomplete
LU preconditioner that can accelerate the convergence of Krylov
subspace iterative solvers. For further information, see
http://crd.lbl.gov/~xiaoye/SuperLU/.

* SuperLU_dist 2.3. SuperLU_dist is a package of parallel,
sparse, direct linear-system solvers (available in Cray LibSci).
For further information, see
http://crd.lbl.gov/~xiaoye/SuperLU/.

* ParMETIS 3.1.1. ParMETIS (Parallel Graph Partitioning and
Fill-reducing Matrix Ordering) is a library of routines that
partition unstructured graphs and meshes and compute fill-
reducing orderings of sparse matrices. For further information,
see
http://glaros.dtc.umn.edu/gkhome/views/metis/.

* HYPRE 2.6.0b. HYPRE is a library of high-performance
preconditioners that use parallel multigrid methods for both
structured and unstructured grid problems (not included with
petsc-complex). For further information, see
http://www.llnl.gov/CASC/linear_solvers/.

* SUNDIALS 2.4.0 (SUite of Nonlinear and DIfferential/ALgebraic
equation Solvers) consists of 5 solvers: CVODE, CVODES, IDA,
IDAS, and KINSOL. In addition, SUNDIALS provides a MATLAB
interface to CVODES, IDAS, and KINSOL that is called sundialsTB.
For further information, see
https://computation.llnl.gov/casc/sundials/main.html.

* Scotch 5.1.1 Scotch is a software package and libraries for
sequential and parallel graph partitioning, static mapping,
sparse matrix block ordering, and sequential mesh and
hypergraph partitioning. For further information, see
http://www.labri.fr/perso/pelegrin/scotch

Documentation:
--------------
http://graal.ens-lyon.fr/MUMPS/
http://crd.lbl.gov/~xiaoye/SuperLU/
http://www.llnl.gov/CASC/linear_solvers/
http://glaros.dtc.umn.edu/gkhome/views/metis/
http://www.labri.fr/perso/pelegrin/scotch/
https://computation.llnl.gov/casc/sundials/main.html

PGI
The trilinos module is not supported with use of PGI/11. This is due to
an incompatibility in c++ code built with previous versions of PGI. For
more information on this incompatibility see the PGI manual at
http://www.pgroup.com/doc/pgiwsrn.pdf. The trilinos module is still
supported with the use of PGI 10.

ATP
- The ATP signal handler (contained in atpSigHandler.o)
is now compiled with position independent code (PIC).
Without this, linking it into shared, dynamic applications
was not possible.

- An ALPS bug can interact with ATP, causing compute nodes
to not free all of their resources during job exit. Such
nodes can become unusable. This is fixed in CLE 3.1 UP02.
However, ATP 1.1.2 has redundantly fixed this situation, for
the benefit of those who have not yet upgraded to CLE 3.1 UP02.