blob: 255a2a20da316a4eb21859d0cbfaec426e42d64b [file] [log] [blame]
NAS Parallel Benchmarks 3.3 - Serial version (NPB3.3-SER)
---------------------------------------------------------
NAS Parallel Benchmarks Team
npb@nas.nasa.gov
This directory contains the serial implementation of the NAS
Parallel Benchmarks, Version 3.3 (NPB3.3-SER). A brief
summary of the new features introduced in this version is
given below.
For changes from different versions, see the Changes.log file
included in the upper directory of this distribution.
For explanation of compilation and running of the benchmarks,
please refer to README.install.
This version (3.3) introduces a new problem size (class E) to seven
of the benchmarks (BT, SP, LU, CG, MG, FT, and EP). The version
also includes the class D problem size for the IS benchmark.
The release is merged with the vector codes for the BT and
LU benchmarks, which can be selected with the VERSION=VEC option
during compilation. However, successful vectorization highly
depends on the compiler used. Some changes to compiler directives
for vectorization in the current codes (see *_vec.f files)
may be required.
Main changes in NPB3.3-SER:
- Introduction of the Class E problem size (except for IS, UA, DC)
- Include the Class D problem size for the IS benchmark.
The "Bucket" option is now the default.
- Merged with the vector codes for the BT and LU benchmarks.
- LU-HP is no longer included in the distribution.
Main changes in NPB3.2-SER:
- Convert C++ version of DC to plain C.
Main changes in NPB3.1-SER:
- Include the Class D problem size in all benchmarks except for
the IS benchmark.
- Redefine the Class W problem size for MG to avoid too fast
convergence. The new size is 128x128x128, 4 iterations.
- Use relative errors for verification in MG and CG, which is
consistent with other benchmarks.
- Include one SSOR iteration before the time step loop in both
LU and LU-HP to touch all pages.
- Include the UA benchmark
- Include the DC benchmark
The serial version of NPB3.0 (NPB3.0-SER) is based on NPB2.3-serial
with the following improvements:
- memory optimization for BT and SP
- two implementations included for LU
- restructured FT