Enabling AMD64 Software Development

1
Enabling AMD64 Software Development

• Ken SchuttStrategic Alliance Manager AMD64
  Tools Infrastructure

2
FORTRAN and C/C Compiler Support

• AMD and The Portland Group Compiler Technology of
  STMicroelectronics are working together to bring
  The Portland Group Compiler Technology to AMD64
• Support will include
• F90 F77
• Some F95 extensions also included
• SPECcpu2000 explicitly supported
• Optimized 32-bit and 64-bit code generation
• Both Linux and Windows will be supported
• Optimized 32-bit support
• Native 64-bit code generation
• OpenMP support
• Full debugging support
• The C/C compilers will leverage The Portland
  Group Compiler Technology for ST's embedded
  processors.
• Beta versions freely available today at
  www.pgroup.com/AMD64
• Production quality in June, 2003 via PGI
  Workstation 5.0 and CDK 5.0

The Portland Group Compiler Technology
3
Compiler Features

• PGI Fortran/C/C Key Features
• PGF90 Fortran 90 Compiler
• Robust, commercially-supported F90 w/ F95
  features
• PGC C Compiler
• EDG 3.0 ISO/ANSI C
• EDG Microsoft Compatibility Switch
• STLport v4.5 Standard Template Library
• PGCC ANSI C Compiler w/ KR support
• OpenMP auto-parallelization for SMP systems
  F77, F90, C, C
• AMD64 specific - More Registers, Scalar SSE (No
  x87 FP), Scheduling, Tuned Intrinsics, Better ABI
• General - SMART Asm Optimizer, SSE/SSE2, x87 FP
  when needed on x86-32, x86-32 ABI, IPA, PFA,
  Cache optimizations (e.g. tiling), Cache-aware
  Vectorizer
• Miscellaneous Fortran byte-swapping I/O Large
  File support
• GNU Interoperability on Linux - cross-link
  gcc/g77 objects/libraries

4
PGI CDK 5.0 - Additional FeaturesTentative
Availability - July, 2003

• Highly Scaleable
• MPI-CH - Pre-configured libraries and utilities
  for ethernet- based x86 and AMD64/Linux
  clusters
• PBS Portable Batch System batch-queuing from
  NASA Ames and MRJ Technologies
• ScaLAPACK - Pre-compiled distributed-memory
  parallel Math Library
• ACML The AMD Core Math Library is scheduled to
  be included
• Training Tutorials (OSC), exercises, examples
  and benchmarks for MPI, OpenMP and HPF
  programming

5
PGDBG Debugger

• OpenMP MPI Parallel PGDBG Debugger
• Debug MPI OpenMP Threaded Programs
• For each or all Processes AND for each or all
  Threads Step, Break, Examine Variables,
  Continue,
• Shared-Memory Debugging
• OpenMP or Automatic Thread-level parallelism
• For each OR for all Threads Step, Break, Halt,
  Windows, Continue,
• Distributed-Memory MPI Debugging
• Debug MPI Processes
• For each OR for all Processes Step, Break,
  Halt, Windows, Continue,
• Hybrid-Memory MPI OpenMP Debugging
• Multiple Nodes where each node has more than one
  SMP CPU
• Control each process control each thread
  separately or as groups.

6
PGI Compiler Technical Support

• Release notes point out known issues/workarounds
• http//www.pgroup.com/beta_docs/pgiwsrn50.pdf
• All technical issues should be sent to PGI
• trs_at_pgroup.com
• This will reach the AMD64 compiler project team
• Please include detailed description of issue
  including machine configuration, OS and compiler
  flags used
• For all high-priority issues email Ken Schutt
  (Iinternal Use Only)
• Ken.schutt_at_amd.com
• I will drive these issues through PGI engineering
• PGI will issues bug fixes/performance
  improvements in the 5.x releases
• PGI is willing to work directly with
  high-priority accounts

7
Optimized Numerical LibrariesACML/libm

• AMD and The Numerical Algorithms Group are
  working together to develop the AMD Core Math
  Library (ACML) for use with mathematical,
  engineering, scientific and financial
  applications as well as general HPC computing
• ACML will be comprised of
• Basic Linear Algebra Subroutines (BLAS) levels 1,
  2 and 3
• A wide variety of Fast Fourier Transforms (FFTs)
• Linear Algebra Package (LAPACK)
• ACML will have the following features
• Fortran and C Interfaces
• Highly optimized routines for the AMD64
  Instruction Set
• Ability to address single-, double-,
  single-complex and double-complex data types
• ACML 0.9 Beta is freely downloadable from
  www.developwithamd.com/acml
• ACML 1.0 is scheduled to be released in June,
  2003
• NAG and AMD are also working together to develop
  highly-optimized numerical routines which will be
  available with commercially available OSes
• Technical support is available via
  acml.support_at_amd.com

8
Other Development Tools Publicly
Announced

• Absoft will be bringing their full set of FORTRAN
  toolsets to the AMD64 architecture on both Linux
  and Windows
• Potential beta testers should send email to
  opteronbeta_at_absoft.com
• MigraTECs source code migration tool, 64Express,
  is now available to aid in the migration of C/C
  code from 32-bit to 64-bit
• MigraTECs cross-platform tool, 32Drect, is now
  available to assist in cross-platform migrations
  (i.e. Solaris to Linux)
• Etnus has announced 32-bit support of x86-64 with
  their TotalView distributed debugging product
• Discussions are underway regarding 64-bit support

9
Other Development Tools Publicly
Announced

• ATLAS (Automatically Tuned Linear Algebra
  Subroutines)
• ATLAS has incorporated optimized 64-bit Linux
  routines to their 3.5.0 Developer release -
  http//math-atlas.sourceforge.net/
• Further 64-bit optimizations are forthcoming
• Compuware has announced support for the SoftICE
  driver debugger and DriverStudio for Win64
• The RC is currently available from Compuware via
  their Beta program
• Scyld Computing has announced their intent to
  support the AMD64 architecture with their Beowulf
  product around time of Opteron launch
• MPICH
• MPICH is available via the open-source community
  and Linux distributions

10
Other Development Tools Publicly
Announced

• Announced 32-bit support with Vampir/Vampirtrace
  for the AMD64 architecture
• 64-bit support discussions are underway
• Announced support for AMD64 with their
  Distributed Debugger Tool (DDT)
• This is the first graphical software debugger to
  support AMD64 with a 64-bit OS
• Blackdown Java
• Announced support that their J2SE Version 1.4.2
  will support the AMD64 architecture on the Linux
  OS
• Blackdown is based on Suns HotSpot technology

11
Other Development Tools Publicly Announced

• - Announced their support for both 32-bit and
  64-bit cluster management software supporting the
  AMD64 Architecture
• - Announced their support for 64-bit versions of
  their popular message passing implementations
  MPIPro (1.2) and ChaMPIon Pro (2.1)
• - Announced the release of the NAGWare F95
  compilers for 64-bit Linux - Available via
  www.nag.com/f95AMD