Dear
Supercomputer User,
This
is to invite you to submit an entry for the new rank list of supercomputers
that provide top application performance. Below is the description
of the new Top Application Performers (TAP) rank list, which was announced
at the SC 2002 conference. In contrast to other rank lists, the TAP
list measures high-performance computers' ability to perform well on
full-scale applications. URL: www.ece.purdue.edu/ParaMount/TAP.
We look forward to seeing your high-performance computer on
TAP!
The TAP team
Purdue University
>>> TAP
50, The 50 Top Application Performers <<<
The
TAP list is a new rank list of high-performance computer systems, based
on real, industrial applications and data sets.
Top Application Performers (TAP) is an effort to encourage
the use of real applications for measuring the performance of high-performance
computer systems. The need for evaluating computer systems with realistic benchmarks
is evident and widely acknowledged. However, so far there has not been an effort
to demonstrate and rank HPC system performance based on such metrics.
TAP 50 will fill this void. It ranks the fastest computer systems
at www.ece.purdue.edu/ParaMount/TAP. TAP complements other computer rank lists,
most notably the TOP 500, which measures HPC systems based on Linpack benchmarks.
Compared to a TOP 500 entry, submitting to the TAP list is more involved. It
requires running large-scale computational applications with large data sets,
adhering to specific benchmark run rules, and preparing a result report, which
is reviewed by benchmarking experts.
The TAP list measures performance based on the benchmark suites
of the High-Performance Group of the Standard Performance Evaluation Corporation
(SPEC). Currently, these suites include the SPEC HPC and the SPEC OMP benchmarks.
The HPC2002 benchmark suite forms the primary metric for the TAP list. It is
based on real high-performance computing (HPC) applications and the MPI and OpenMP
standards for parallel processing. Hence it represents state-of-the-art HPC practices
most realistically. In addition, TAP presents a rank list based on SPEC/HPG's
OpenMP benchmarks. These codes consist of a number of computation-intensive applications
that have been transformed by SPEC into OpenMP parallel form. They are especially
suited to measure shared-memory parallel computers.
Generating
a TAP Submission:
In order to generate a TAP entry, you need to run the SPEC
HPC or the SPEC OMP benchmarks using the same run rules as for generating valid
SPEC results. You will submit results through the same channels as if submitting
SPEC results. The rules are described in the benchmark distribution as well as
on the web page of SPEC's High-Performance Group www.spec.org/hpg/. The submitted
result will be screened like any SPEC result submission. The submitter must be
a member of the SPEC High-Performance Group. SPEC membership is open to both
industrial and nonprofit organizations. Non-members can find a sponsor to submit
the results. This can either be the vendor company of the system being benchmarked
or an academic SPEC member. Purdue University considers sponsoring academic participants;
send mail to topap@ecn.purdue.edu.
Ranking:
Each TAP list has its own metric for ranking. For example,
for the newest TAP list, the benchmark SPEC HPC 2002 with the medium data set
is used. For each list entry, the results of the different (currently three)
benchmarks are combined into one number by weighting the SPEC metric by a reference
run and summing the numbers from all the benchmarks for that data set.
How to Obtain
More Information:
For more information about the benchmarks: http://www.spec.org/hpg/.
To obtain benchmark codes and for SPEC membership, contact: info@spec.org. For
general information about TAP: http://www.ece.purdue.edu/ParaMount/TAP or send
mail to: topap@ecn.purdue.edu |