ARSC T3D Users' Newsletter 47, August 11, 1995

The 1.2.2 Release of the Programming Environment (cont.)

Currently at ARSC we are testing the 1.2.2 Programming release. Complete directions on how to use this new release can be found by executing:

  news compilers
on denali. There are lots of new features in this release. Below is the full text of the release letter. If there are any problems with using this release please contact Mike Ess.

  > Programming Environment 1.2.2 Release Letter
  > 1.0        Introduction
  > ---------------------
  > This release letter describes the Programming Environment 1.2.2 releases for
  > Cray Parallel Vector Processor (PVP) and Cray Massively Parallel Processor
  > (MPP) systems. The major focus for these releases is performance enhancements
  > and the addition of new library routines for Cray MPP systems. Bugfixes were
  > made in most products across all platforms.
  > Users will see the following enhancements with the Programming Environment
  > 1.2.2 releases for Cray MPP systems:
  >   o - Increased performance from the CF77, CF90, and Cray Standard C (SCC)
  >       compilers.
  >   o - Addition of array intrinsics that perform segmented scan and parallel
  >       prefix operations.
  >   o - Included in the Cray Research scientific library, libsci, is new software
  >       from LAPACK and ScaLAPACK. The ScaLAPACK software consists of new
  >       parallel 2-D and 3-D Fast Fourier Transform (FFT) routines, and 
  >       performance enhancements to FFT routines and single-processor level 3
  >       Basic Linear Algebraic Subroutines (BLAS3).
  > 2.0        Programming environments for Cray PVP and MPP systems
  > ---------------------
  > The following Cray programming environment are included in the 1.2.2 releases:
  >   o -  CF77 6.2.2 Programming Environment for both Cray PVP and MPP systems
  >   o -  CF90 Programming Environment 1.2.2 for both Cray PVP and MPP systems
  >   o -  Cray Standard C Programming Environment 1.2.2 for both Cray PVP and
  >        MPP systems
  > The programming environments provide an integrated development environment and
  > combine compilers with development tools and libraries. Each of the programming
  > environments includes the CrayTools 1.3.2 package, the CrayLibs 1.2.2 package
  > for either Cray PVP and MPP systems, and one of the following compilers:
  >   o -  CF77 6.2.1 for Cray MPP systems (CF77_M 6.2.1)
  >   o -  CF77 6.0.4 for Cray PVP systems *
  >   o -  CF90 0.1.2 compiler for Cray MPP system (CF90_M 0.1.2)
  >   o -  CF90 1.0.3 for Cray PVP systems
  >   o -  Cray Standard C 4.0.4 for Cray MPP systems (SCC_M 4.0.4)
  >   o -  Cray Standard C 4.0.3 for Cray PVP systems *
  > *Note: CF77 6.0.4 and Cray Standard C 4.0.3 were also in the Programming
  >        Environment 1.2. release packages and remain unchanged from those
  >        releases.
  > You cannot order programming environment components separately; you must order
  > a complete programming environment. To order a programming environment, contact
  > your Cray Research service representative or complete a software order form in
  > the CRInform service.
  > 3.0        Hardware and operating system dependencies
  > ---------------------
  > The Programming Environment 1.2.2 products for Cray MPP systems are supported
  > on the CRAY T3D series running UNICOS MAX, release or later, and
  > UNICOS, release 8.0 or later. 
  > The Programming Environment 1.2.2 products are supported on the following
  > Cray PVP systems running the UNICOS operating system, release 8.0 or later: 
  >   o -  CRAY Y-MP EL series
  >   o -  CRAY C90 and CRAY C90D series
  >   o -  CRAY J90 series
  >   o -  CRAY T90 series
  >   o -  CRAY Y-MP, CRAY Y-MP E and CRAY Y-MP M90 series.
  > 4.0        Dependencies for Cray MPP systems
  > ---------------------
  > The CF77_M 6.2.1 compiling system and CrayLibs_M 1.2.2 must be installed at
  > the same time due to dependencies between these products in the parallel
  > prefix and segmented scan shared array intrinsics: PREMIN, PREMAX, PRESUM,
  > PREPROD, SCANMIN, SCANMAX, SCANSUM, and SCANPROD. These array intrinsics are
  > being introduced for the first time in this release. If programs using these
  > array intrinsics are compiled with CF77_M 6.2.1, CrayLibs_M 1.2.2 (or later)
  > must also be used.
  > 5.0        Compatibilities and Differences
  > ---------------------
  > The user interface for loop unrolling in the CF77_M compiling system has
  > changed. In previous releases, when a user specified the following
  >        CDIR $ UNROLL [4]
  > 5 copies of the loop body were created; the original plus 4. This was not
  > compatible with the CF77 compiling system for Cray PVP systems implementation.
  > To achieve compatibility, with this release the example above results in a
  > total of 4 copies of the loop body, the original plus 3. 
  > 6.0        Software features overview
  > ---------------------
  > This section describes enhancements made to the Cray MPP systems programming
  > environments for these releases.
  > 6.0.1         Features for Cray MPP systems
  > ---------------------
  > For MPP systems, the programming environment 1.2.2 releases contain the
  > following enhancements:
  >   o -  Performance in the CF77_M compiling system was enhanced by the addition
  >        of optimizations that improve cache performance for stack data.
  >   o -  The CF77_M compiling system, CF90_M compiler and SCC_M compiler enhanced
  >        performance through optimizations that stream-line inner loop
  >        computations.
  >   o -  The addition of SHMEM_MY_PE() and SHMEM_N_PES(), functions that return
  >        the number of the PE executing and the number of PEs available to the
  >        program, respectively. These functions are especially useful with the
  >        CF90 compiler where intrinsics for this purpose are not available.
  >   o -  Performance improvements in the library routine, SHMEM_GET, occur for
  >        transfer sizes of 1-3 words. Cray Research test suites saw up to 1.86
  >        times improvement.
  >   o -  The Parallel Virtual Machine (PVM) global operation function,
  >        pvm_reduce() has improved performance. pvm_reduce() was previously
  >        layered upon pvm_psend() and pvm_precv() functions and used the same
  >        linear algorithm as is used in the public domain version of PVM. The
  >        optimized version uses direct shared memory and achieves more 
  >        logarithmic scaling.
  >   o -  The following Fortran shared array intrinsics were added: PRESUM,
  >        PREPROD, PREMIN, and PREMAX, (parallel prefix operations) and SCANSUM,
  >        SCANPROD, SCANMIN, and SCANMAX (segmented scan operations). These
  >        intrinsic functions are variations on traditional reduction functions.
  >   o -  Enhancements to the Libsci libraries for the Cray MPP systems include
  >        the following:
  >        1) The addition of an optimized implementation of ScaLAPACK, a set of
  >           scalable algorithms for solving linear algebra problems. ScaLAPACK 
  >           contains a number of linear solvers, similarity and equivalence
  >           reductors, and an Eigensystem solver for real symmetric matrices.
  >           These algorithms use the Basic Linear Algebra Communication Sub-
  >           programs (BLACS) for communication and the Parallel Basic Linear
  >           Algebraic Subroutines (PBLAS) for computation. The PBLAS, new in
  >           this release, provide distributed-memory software comparable to
  >           the BLAS in shared-memory codes.
  >        2) New FFT routines were added to perform parallel 2-D and 3-D
  >           real-to-complex and complex-to-real FFTs. 
  >        3) The performance of the FFT subroutine, CCFFT, has been improved for
  >           prime-length vectors, and for lengths that are not factorizable into
  >           powers of 2, 3, and 5. The single-PE routines SCFFT and CSFFT have
  >           been extended to work with odd-length vectors.
  >        4) Newly optimized versions of the following routines were added:
  >           - SNRM2, SCNRM2, SROTG, and CROTG from the Level 1 BLAS routines.
  >           - STRSM, STRMM, and the other Level 3 BLAS routines.
  > 6.0.2         Features for Cray PVP and MPP systems
  > ---------------------
  > The libsci LAPACK libraries added support on both Cray PVP and MPP systems for
  > factorization variant QR with pivoting, linear least squares driver routines,
  > and reductions to condensed form for eigenvalue computations. These libraries
  > also contain performance improvements to xLARTG and xLARFG, the auxiliary
  > routines that generate elementary transformations. LAPACK is a package of
  > software for solving linear systems and matrix eigenvalue problems. 
  > 7.0        Software release documentation
  > ---------------------
  > The User Publications Catalog, publication CP-0099, describes Cray Research
  > product documentation available to customers. Cray Research customers who
  > subscribe to Cray Inform (CRInform) can access this information on the CRInform
  > system and can use CRInform to order publications.
  > 7.0.1         Programming Environment 1.2.2 updated documents
  > ---------------------
  > In addition to the numerous man pages updated for this release, the following
  > manuals have also been updated:
  >    - Cray MPP Fortran Reference Manual, publication SR-2504 6.2.2 (available
  >       online in Docview format only)
  >    - Programming Environment Installation Guide, publication SG-5217 1.2.2
  >      (available in printed format only)
  >         CrayDoc
  > ---------------------
  > The CF90, CF77 and Cray Standard C programming environment documentation can
  > be viewed with the CrayDoc online documentation reader. CrayDoc, which runs
  > on workstations, gives you online access to compiler and tools manuals. 
  > A separate CrayDoc disk accompanies the software for each product. There is
  > no separate charge for the disk that accompanies the release packages; if you
  > order additional disks you will be charged an additional fee. Contact your
  > Cray Research representative for information on how to order CrayDoc.
  >         Docview
  > ---------------------
  > The Docview program provides online access to a library of printed Cray
  > Research documents and local documents supplied by your site. You can use
  > the Docview program to find a particular piece of information in a publication
  > quickly or to search through the entire set of available documents for
  > information associated with a topic you specify. You can view the information
  > online or you can write it to disk to print a hard copy. For more information
  > on the Docview utility, for systems running UNICOS 7.0, see the Docview User's
  > Guide, publication SG-2109, and the Docview Writer's Guide, publication SG-2118;
  > for systems running UNICOS 8.0, see the User's Guide to Online Information,
  > publication SG-2143. 
  > Programming Environment 2.0 and other future programming environment releases
  > will not include manuals in Docview format.The Docview program will go into
  > maintenance mode with the UNICOS 9.0 release. After that point Cray Research
  > will no longer provide Docview versions of the manuals. 
  >         Release Letter
  > ---------------------
  > You can access this release letter electronically; ASCII and PostScript files
  > are available on the Cray Research CRInform system. 

IEEE Document

In preparation for the Cray T90 with IEEE arithmetic, CRI has produced a preliminary draft of manual SN-2194 2.0 Migration to CRAY T90 IEEE Floating-Point. The IEEE implementation on the T3D is different than that on the T90, but there is good information for T3D users in this document like:
  1. Explanation of IEEE rationale and conformance
  2. Examples of data conversion (through ASSIGN)
  3. Description of code migration from Y-MP/C90 to T90
  4. A Summary of T3D/T3E IEEE conformance (Appendix A)
Contact Mike Ess about getting a copy of this large postscript document.

List of Differences Between T3D and Y-MP

The current list of differences between the T3D and the Y-MP is:
  1. Data type sizes are not the same (Newsletter #5)
  2. Uninitialized variables are different (Newsletter #6)
  3. The effect of the -a static compiler switch (Newsletter #7)
  4. There is no GETENV on the T3D (Newsletter #8)
  5. Missing routine SMACH on T3D (Newsletter #9)
  6. Different Arithmetics (Newsletter #9)
  7. Different clock granularities for gettimeofday (Newsletter #11)
  8. Restrictions on record length for direct I/O files (Newsletter #19)
  9. Implied DO loop is not "vectorized" on the T3D (Newsletter #20)
  10. Missing Linpack and Eispack routines in libsci (Newsletter #25)
  11. F90 manual for Y-MP, no manual for T3D (Newsletter #31)
  12. RANF() and its manpage differ between machines (Newsletter #37)
  13. CRAY2IEG is available only on the Y-MP (Newsletter #40)
  14. Missing sort routines on the T3D (Newsletter #41)
I encourage users to e-mail in differences that they have found, so we all can benefit from each other's experience.
Current Editors:
Ed Kornkven ARSC HPC Specialist ph: 907-450-8669
Kate Hedstrom ARSC Oceanographic Specialist ph: 907-450-8678
Arctic Region Supercomputing Center
University of Alaska Fairbanks
PO Box 756020
Fairbanks AK 99775-6020
E-mail Subscriptions: Archives:
    Back issues of the ASCII e-mail edition of the ARSC T3D/T3E/HPC Users' Newsletter are available by request. Please contact the editors.
Back to Top