dolfin team mailing list archive
-
dolfin team
-
Mailing list archive
-
Message #15405
[HG DOLFIN] Add first version of script for testing parallel speedup. Run with ./bench.
One or more new changesets pushed to the primary dolfin repository.
A short summary of the last three changesets is included below.
changeset: 7067:5efbfb16a1b8aab6be985d739d085a8b73afb707
tag: tip
user: Anders Logg <logg@xxxxxxxxx>
date: Tue Sep 15 14:04:39 2009 +0200
files: bench/fem/speedup/Poisson.h bench/fem/speedup/Poisson.ufl bench/fem/speedup/SConstruct bench/fem/speedup/assemble-poisson.cpp bench/fem/speedup/bench bench/fem/speedup/bench.log bench/fem/speedup/run-bench bench/fem/speedup/solve-poisson.cpp
description:
Add first version of script for testing parallel speedup. Run with ./bench.
Have not tried on a big problem or a machine with many processors.
Example results from my dual-core laptop on a very small problem:
Speedup | Assemble Assemble + solve
--------------------------------------
1 | 1 1
2 | 1 0.75439
3 | 0.83333 0.54386
4 | 0.83333 0.5614
5 | 0.5 0.54386
6 | 0.5 0.57895
7 | 0.33333 0.66667
8 | 0.33333 0.64912
(Negative speedups)
changeset: 7066:a59e1c1e9d86f373c6289ed7c7a76c03c3a62c9f
user: Anders Logg <logg@xxxxxxxxx>
date: Tue Sep 15 13:57:07 2009 +0200
files: dolfin/log/Logger.cpp dolfin/log/Table.cpp dolfin/log/Table.h dolfin/mesh/Mesh.cpp dolfin/parameter/Parameters.cpp
description:
Bug fix for str() function in Table, now working with info() in Python
changeset: 7065:a54e8e47dfe0fd5c6601da9f9aafe265845b934c
user: Anders Logg <logg@xxxxxxxxx>
date: Tue Sep 15 11:04:51 2009 +0200
files: demo/fem/simple/cpp/main.cpp demo/fem/simple/python/demo.py
description:
Add not_working_in_parallel to demo/fem/simple. Fails when trying to partition
a mesh consisting of a single cell. (There should be a better error message than
a segmentation fault.)
----------------------------------------------------------------------
For more details, visit http://www.fenics.org/hg/dolfin