xfygogo/phasicFlow
1
0
Fork 0
mirror of https://github.com/PhasicFlow/phasicFlow.git synced 2025-06-12 16:26:23 +00:00
Code Issues Packages Projects Releases Wiki Activity

Zoltan is added as thirdParty package

Browse Source
This commit is contained in:
Hamidreza
2025-05-15 21:58:43 +03:30
parent 83a6e4baa1
commit d7479cf1bd
3392 changed files with 318142 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

127
thirdParty/Zoltan/docs/ug_html/ug_graph_vs_hg.html vendored Normal file
View File

@ -0,0 +1,127 @@
<!-------- @HEADER
!
! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
! Zoltan Toolkit for Load-balancing, Partitioning, Ordering and Coloring
! Copyright 2012 Sandia Corporation
!
! Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
! the U.S. Government retains certain rights in this software.
!
! Redistribution and use in source and binary forms, with or without
! modification, are permitted provided that the following conditions are
! met:
!
! 1. Redistributions of source code must retain the above copyright
! notice, this list of conditions and the following disclaimer.
!
! 2. Redistributions in binary form must reproduce the above copyright
! notice, this list of conditions and the following disclaimer in the
! documentation and/or other materials provided with the distribution.
!
! 3. Neither the name of the Corporation nor the names of the
! contributors may be used to endorse or promote products derived from
! this software without specific prior written permission.
!
! THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
! EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
! IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
! PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
! CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
! EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
! PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
! PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
! LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
! NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
! SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
!
! Questions? Contact Karen Devine kddevin@sandia.gov
! Erik Boman egboman@sandia.gov
!
! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
! @HEADER
------->
<!DOCTYPE html PUBLIC "-//w3c//dtd html 4.0 transitional//en">
<html>
<head>
<meta http-equiv="Content-Type"
content="text/html; charset=iso-8859-1">
<meta name="GENERATOR"
content="Mozilla/4.76 [en] (X11; U; Linux 2.4.2-2smp i686) [Netscape]">
<meta name="sandia.approval_type" content="formal">
<meta name="sandia.approved" content="SAND2007-4748W">
<meta name="author" content="Zoltan PI">
<title>Zoltan User's Guide: Graph vs. Hypergraph Partitioning</title>
</head>
<body bgcolor="#ffffff">
<div align="right"><b><i><a href="ug.html">Zoltan User's Guide</a>&nbsp; |
<a href="ug_alg_parmetis.html"> Next</a>&nbsp; |
&nbsp;<a href="ug_alg_graph.html">Previous</a>&nbsp;
</i>
</b></div>
<h2>
<a name="Graph vs Hypergraph Partitioning">
</a>Graph vs Hypergraph Partitioning</h2>
Graph partitioning has proven quite useful in scientific computing.
<a href="ug_alg_hypergraph.html">Hypergraph partitioning</a>
is a more recent improvement
that uses a hypergraph model, which is
often a more accurate model than the graph model for scientific computing.
(Hypergraphs contain hyperedges which connect two <i>or more</i>
vertices.)
See <a href="ug_refs.html#catalyurek99">[Catalyurek & Aykanat]</a> and
<a href="ug_refs.html#hendrickson-kolda">[Hendrickson & Kolda]</a>
for further details.
You do not need to understand the underlying models
to use graph or hypergraph partitioning for load-balancing
in Zoltan. The basic trade-offs are:
<ul>
<li> Hypergraph partitioning usually produces partitions (assignments)
of higher quality than graph partitioning, which may reduce
communication time in parallel applications (up to 30-40%
reduction has been reported). However, hypergraph
partitioning takes longer time to compute. </li>
<li> The graph model is restricted to symmetric data dependencies.
If you have a non-symmetric problem, we recommend hypergraph
partitioning. </li>
</ul>
<h3>Migrating from ParMetis to PHG in Zoltan</h3>
If you already use Zoltan for graph partitioning (via ParMetis),
there are three ways to switch to the Zoltan-PHG hypergraph partitioner:
<ol>
<li>The quick and easy way: Just change the <a href="ug_alg.html#LB_METHOD">
LB_METHOD</a> to "Hypergraph".
Zoltan will then use the graph query functions (presumably
already implemented) to construct a hypergraph model, which
is similar to but not equivalent to the graph.
<li>The proper way: Change the LB_METHOD, but also implement and
register the <a href="ug_query_lb.html#ZOLTAN_HG_SIZE_CS_FN">
hypergraph query functions</a>
required by Zoltan. These may give a more accurate representation
of data dependencies (and communication requirements) for your
application.
<li>If you really want graph (not hypergraph) partitioning:
Just change the <a href="ug_alg.html#LB_METHOD">
LB_METHOD</a> to "Graph".
Zoltan will then use PHG as a graph partitioner, which is slower than
ParMetis but often produces better partitions (lower cuts).
</ol>
Technical note: A hypergraph is constructed from the graph as follows:
The vertices are the same in the hypergraph as in the graph. For each vertex
<i>v</i>, create a hyperedge that consists of all neighbors in the graph
and <i>v </i> itself.
<hr width="100%">[<a href="ug.html">Table of Contents</a>&nbsp; |
<a href="ug_alg_parmetis.html"> Next:&nbsp; ParMETIS</a>&nbsp; |
&nbsp; <a href="ug_alg_graph.html">Previous:&nbsp;
Graph Partitioning</a>&nbsp; |&nbsp; <a href="https://www.sandia.gov/general/privacy-security/index.html">Privacy and Security</a>]
</body>
</html>