[176] | 1 | <?xml version="1.0" encoding="UTF-8"?> |
---|
| 2 | <!-- |
---|
| 3 | Copyright 2002-2004 The Apache Software Foundation |
---|
| 4 | |
---|
| 5 | Licensed under the Apache License, Version 2.0 (the "License"); |
---|
| 6 | you may not use this file except in compliance with the License. |
---|
| 7 | You may obtain a copy of the License at |
---|
| 8 | |
---|
| 9 | http://www.apache.org/licenses/LICENSE-2.0 |
---|
| 10 | |
---|
| 11 | Unless required by applicable law or agreed to in writing, software |
---|
| 12 | distributed under the License is distributed on an "AS IS" BASIS, |
---|
| 13 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
---|
| 14 | See the License for the specific language governing permissions and |
---|
| 15 | limitations under the License. |
---|
| 16 | --> |
---|
| 17 | <!DOCTYPE document PUBLIC "-//APACHE//DTD Documentation V2.0//EN" "http://forrest.apache.org/dtd/document-v20.dtd"> |
---|
| 18 | <document> |
---|
| 19 | <header> |
---|
| 20 | <title>The Hoard Multiprocessor Memory Allocator</title> |
---|
| 21 | </header> |
---|
| 22 | |
---|
| 23 | <!-- |
---|
| 24 | <articleinfo> |
---|
| 25 | <author> |
---|
| 26 | <firstname>Emery</firstname> |
---|
| 27 | <surname>Berger</surname> |
---|
| 28 | <affiliation>University of Massachusetts Amherst</affiliation> |
---|
| 29 | <street>Department of Computer Science</street> |
---|
| 30 | <city>Amherst</city> |
---|
| 31 | <state>Massachusetts</state> |
---|
| 32 | <country>USA</country> |
---|
| 33 | <email>emery@cs.umass.edu</email> |
---|
| 34 | </author> |
---|
| 35 | <pubdate>2004-12-08</pubdate> |
---|
| 36 | <copyright> |
---|
| 37 | <year>2004</year> |
---|
| 38 | <holder role="mailto:emery@cs.umass.edu">Emery Berger</holder> |
---|
| 39 | </copyright> |
---|
| 40 | <abstract> |
---|
| 41 | Documentation for the Hoard scalable memory allocator, including build and usage directions for several platforms. |
---|
| 42 | </abstract> |
---|
| 43 | </articleinfo> |
---|
| 44 | --> |
---|
| 45 | |
---|
| 46 | <body> |
---|
| 47 | |
---|
| 48 | <p class="quote"> |
---|
| 49 | ...if you'll be running on multiprocessor machines, ... <strong>use <a href="http://www.cs.umass.edu/~emery">Emery Berger</a>'s excellent Hoard multiprocessor memory management code</strong>. It's a drop-in replacement for the C and C++ memory routines and is very fast on multiprocessor machines. |
---|
| 50 | <em><a href="http://www.nerdbooks.com/item.php?id=0735615365">Debugging Applications for Microsoft .NET and Microsoft Windows</a>, Microsoft Press, 2003</em> |
---|
| 51 | </p> |
---|
| 52 | |
---|
| 53 | <!-- |
---|
| 54 | <p class="quote"> |
---|
| 55 | <strong>hoard:</strong> |
---|
| 56 | To amass and put away (anything valuable) for preservation, security, |
---|
| 57 | or future use; to treasure up: esp. money or wealth. |
---|
| 58 | <em>Oxford English Dictionary</em> |
---|
| 59 | </p> |
---|
| 60 | --> |
---|
| 61 | |
---|
| 62 | <p> |
---|
| 63 | The Hoard memory allocator is a fast, scalable, and memory-efficient |
---|
| 64 | memory allocator for shared-memory multiprocessors. It runs on a |
---|
| 65 | variety of platforms, including Linux, Solaris, and Windows. Hoard is |
---|
| 66 | a drop-in replacement for malloc(), etc. No change to your source is |
---|
| 67 | necessary. Just link it in or set just one environment variable (see |
---|
| 68 | <a href="using.html">Using Hoard</a> for more |
---|
| 69 | information). <strong>Hoard can dramatically improve the performance |
---|
| 70 | of multithreaded programs running on multiprocessors.</strong> |
---|
| 71 | |
---|
| 72 | |
---|
| 73 | </p> |
---|
| 74 | |
---|
| 75 | <section> |
---|
| 76 | <title>Why Hoard?</title> |
---|
| 77 | |
---|
| 78 | <p> |
---|
| 79 | There are a number of problems with existing memory allocators that make Hoard a better choice. |
---|
| 80 | </p> |
---|
| 81 | |
---|
| 82 | <section> |
---|
| 83 | <title>Contention</title> |
---|
| 84 | <p> |
---|
| 85 | Multithreaded programs often do not scale because the heap is a |
---|
| 86 | bottleneck. When multiple threads simultaneously allocate or |
---|
| 87 | deallocate memory from the allocator, the allocator will serialize |
---|
| 88 | them. Programs making intensive use of the allocator actually slow |
---|
| 89 | down as the number of processors increases. Your program may be |
---|
| 90 | allocation-intensive without you realizing it, for instance, if your |
---|
| 91 | program makes many calls to the C++ Standard Template Library (STL). |
---|
| 92 | </p> |
---|
| 93 | </section> |
---|
| 94 | <section> |
---|
| 95 | <title>False Sharing</title> |
---|
| 96 | <p> |
---|
| 97 | The allocator can cause other problems for multithreaded code. It can |
---|
| 98 | lead to <em>false sharing</em> in your application: |
---|
| 99 | threads on different CPUs can end up with memory in the same cache |
---|
| 100 | line, or chunk of memory. Accessing these falsely-shared cache lines |
---|
| 101 | is hundreds of times slower than accessing unshared cache lines. |
---|
| 102 | </p> |
---|
| 103 | </section> |
---|
| 104 | <section> |
---|
| 105 | <title>Blowup</title> |
---|
| 106 | <p> |
---|
| 107 | Multithreaded programs can also lead the allocator to blowup memory |
---|
| 108 | consumption. This effect can multiply the amount of memory needed to |
---|
| 109 | run your application by the number of CPUs on your machine: four CPUs |
---|
| 110 | could mean that you need four times as much memory. Hoard is a fast |
---|
| 111 | allocator that solves all of these problems. |
---|
| 112 | </p> |
---|
| 113 | </section> |
---|
| 114 | </section> |
---|
| 115 | |
---|
| 116 | <section> |
---|
| 117 | <title>Press</title> |
---|
| 118 | |
---|
| 119 | <p> |
---|
| 120 | <a href="http://www.intel.com/cd/ids/developer/asmo-na/eng/dc/xeon/43893.htm?page=4"><strong>Intel</strong> highlights the benefits of using Hoard</a> (a previous, slower version) on a 4-way Xeon system. |
---|
| 121 | </p> |
---|
| 122 | |
---|
| 123 | <p> |
---|
| 124 | <a href="http://developers.sun.com/solaris/articles/multiproc/multiproc.html"><strong>Sun</strong> concludes that Hoard is more space-efficient</a> than their own allocators. |
---|
| 125 | </p> |
---|
| 126 | |
---|
| 127 | </section> |
---|
| 128 | |
---|
| 129 | <!-- |
---|
| 130 | <section> |
---|
| 131 | <title>How Do I Use Hoard?</title> |
---|
| 132 | <p> |
---|
| 133 | Hoard is a drop-in replacement for malloc(), etc. No change to your source is necessary. Just link it in or set just one environment variable. See <a href="using.html">Using Hoard</a> for more information. |
---|
| 134 | </p> |
---|
| 135 | </section> |
---|
| 136 | --> |
---|
| 137 | |
---|
| 138 | |
---|
| 139 | <section> |
---|
| 140 | <title>Who's Using Hoard?</title> |
---|
| 141 | <p> |
---|
| 142 | Companies using Hoard in their products and servers include <a href="http://www.aol.com">AOL</a>, <a href="http://www.bt.com">British Telecom</a>, <a href="http://www.businessobjects.com">Business Objects</a> |
---|
| 143 | (formerly Crystal Decisions), <a href="http://www.entrust.com">Entrust</a>, <a href="http://www.novell.com">Novell</a>, <a href="http://www.openwave.com">OpenWave Systems</a> (for their |
---|
| 144 | Typhoon and Twister servers), and <a href="http://www.reuters.com">Reuters</a>. |
---|
| 145 | </p> |
---|
| 146 | |
---|
| 147 | <p> |
---|
| 148 | Open source projects using Hoard include the Bayonne GNU telephony |
---|
| 149 | server, the <a href="http://supertech.lcs.mit.edu/cilk/">Cilk</a> parallel |
---|
| 150 | programming language, the <a href="http://www.cs.dartmouth.edu/research/DaSSF/index.html">Dartmouth |
---|
| 151 | Scalable Simulation Framework</a>, and the <a href="http://www.gnu.org/software/commoncpp/">GNU Common C++</a> |
---|
| 152 | system. |
---|
| 153 | </p> |
---|
| 154 | |
---|
| 155 | <p> |
---|
| 156 | Hoard is also a part of several major Linux distributions, including Debian and Novell's SuSe. |
---|
| 157 | </p> |
---|
| 158 | </section> |
---|
| 159 | |
---|
| 160 | |
---|
| 161 | <section> |
---|
| 162 | <title>More Information</title> |
---|
| 163 | <p> |
---|
| 164 | The first place to look for Hoard-related information is at the Hoard |
---|
| 165 | web page, <a href="http://www.hoard.org">www.hoard.org</a>. |
---|
| 166 | </p> |
---|
| 167 | |
---|
| 168 | <p> |
---|
| 169 | There are two mailing lists you should join if you are a |
---|
| 170 | user of Hoard. If you are just interested in being informed of new |
---|
| 171 | releases, join the <a |
---|
| 172 | href="http://groups.yahoo.com/group/hoard-announce/">Hoard-Announce</a> |
---|
| 173 | list. For general Hoard discussion, join the <a |
---|
| 174 | href="http://groups.yahoo.com/group/hoard/">Hoard</a> mailing |
---|
| 175 | list. You can also search the archives of these lists. |
---|
| 176 | </p> |
---|
| 177 | </section> |
---|
| 178 | |
---|
| 179 | |
---|
| 180 | <section> |
---|
| 181 | <title>Technical Information</title> |
---|
| 182 | <p> |
---|
| 183 | For technical details of a previous version of Hoard, read <a href="http://www.cs.umass.edu/~emery/hoard/asplos2000.pdf">Hoard: A Scalable Memory Allocator for Multithreaded Applications</a>, by Emery D. Berger, Kathryn S. McKinley, Robert D. Blumofe, and Paul R. Wilson. The Ninth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS-IX). Cambridge, MA, November 2000. |
---|
| 184 | </p> |
---|
| 185 | </section> |
---|
| 186 | |
---|
| 187 | </body> |
---|
| 188 | </document> |
---|