The Application Perspective for Scalable Data and Task Parallel Languages
HPF and HPC++
This was a talk given by Geoffrey Fox at the Arpa High Performance
Computing and Communication Symposium in Washington D.C. March 17,1994.
This discusses National and Grand Challenges from both Industry and
Academia in terms of the support needed and supplied by HPF, HPC++ and
message passing(MPI) levels. Current data-parallel language standards are
sufficient for efficient implementation of some 30 to 50% of large scale
scientific and engineering simulations. Ongoing research and development
will give us scalable high level support for over 90% of such
applications. Further these application considerations show how to extend
Fortran and C++ parallel constructs to other languages (e.g.ADA) and
intermediate representations such as ANDF.
The slides (Click on number for slide, or text for descriptions):
-
.1=..Title slide: The application perspective for Scalable Data and Task Parallel Languages HPF and HPC++
-
.2=..Importance of HPF, HPC++, to users.
-
.3=..Some specific HPF stories - I.
-
.4=..Some specific HPF stories- II.
-
.5=..Goal of high performance FORTRAN.
-
.6=..General applicability of HPF, HPF+, and HPC++ in four broad problem classes.
-
.7=..Definition of five basic problem classes.
-
.8=..Agile manufacturing as a metaproblem example.
-
.9=..Multidisciplinary analysis and design as the mapping of heterogeneous metaproblems onto heterogeneous metacomputer systems.
-
10=..Further examples of metaproblems.
-
11=..Another dual-use metaproblem: BMC3IS - Decision support.
-
12=..Heterogeneous metaproblem structure for command and control.
-
13=..Three forms of parallelism that any complete programming environment must handle.
-
14=..Features of HPF++: Extensions to HPF.
-
15=..Relevance of HPF and HPC++ features to other languages - I.
-
16=..Relevance of HPF and HPC++ features to other languages - II.
-
17=..Preliminary summary HPF features for different algorithms and generic application methodologies - I.
-
18=..Preliminary summary HPF features for different applications and generic application methodologies - II.
-
19=..Preliminary summary HPF features for different applications and generic application methodologies - III.
-
20=..Preliminary summary HPF features for different applications and generic application methodologies - IV.
-
21=..Overview of industrial and academic applications of HPCC.
-
22=..Industrial and academic applications of HPCC - simulation opportunities I.
-
23=..Industrial and academic applications of HPCC - simulation opportunities II.
-
24=..Industrial and academic applications of HPCC - simulation opportunities III.
-
25=..Industrial and academic applications of HPCC - Information analysis or "data mining".
InfoWarehouse Management team