1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
|
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
<meta name="generator" content="hevea 2.25">
<style type="text/css">
.c000{font-family:monospace}
.li-itemize{margin:1ex 0ex;}
.li-enumerate{margin:1ex 0ex;}
.footnotetext{margin:0ex; padding:0ex;}
div.footnotetext P{margin:0px; text-indent:1em;}
.thefootnotes{text-align:left;margin:0ex;}
.dt-thefootnotes{margin:0em;}
.dd-thefootnotes{margin:0em 0em 0em 2em;}
.footnoterule{margin:1em auto 1em 0px;width:50%;}
.title{margin:2ex auto;text-align:center}
.titlemain{margin:1ex 2ex 2ex 1ex;}
.titlerest{margin:0ex 2ex;}
div table{margin-left:inherit;margin-right:inherit;margin-bottom:2px;margin-top:2px}
td table{margin:auto;}
table{border-collapse:collapse;}
td{padding:0;}
pre{text-align:left;margin-left:0ex;margin-right:auto;}
blockquote{margin-left:4ex;margin-right:4ex;text-align:left;}
td p{margin:0px;}
</style>
<title>DARPA/DOE HPC Challenge Benchmark version 1.5.0beta
</title>
</head>
<body>
<!--HEVEA command line is: hevea -fix -O README.tex -->
<!--CUT STYLE article--><!--CUT DEF section 1 --><table class="title"><tr><td style="padding:1ex"><h1 class="titlemain">DARPA/DOE HPC Challenge Benchmark version 1.5.0beta</h1><h3 class="titlerest">Piotr Luszczek<sup><a id="text1" href="#note1">*</a></sup></h3><h3 class="titlerest">October 12, 2012</h3></td></tr>
</table>
<!--TOC section id="sec1" Introduction-->
<h2 class="section" id="sec1">1  Introduction</h2><!--SEC END --><p>
This is a suite of benchmarks that measure performance of processor,
memory subsytem, and the interconnect. For details refer to the
HPC Challenge web site (<span class="c000">http://icl.cs.utk.edu/hpcc/</span>.)</p><p>In essence, HPC Challenge consists of a number of tests each
of which measures performance of a different aspect of the system.</p><p>If you are familiar with the High Performance Linpack (HPL) benchmark
code (see the HPL web site:
<span class="c000">http://www.netlib.org/benchmark/hpl/</span>) then you can reuse the
build script file (input for <span class="c000">make(1)</span> command) and the input
file that you already have for HPL. The HPC Challenge benchmark
includes HPL and uses its build script and input files with only
slight modifications. The most important change must be done to the
line that sets the <span class="c000">TOPdir</span> variable. For HPC Challenge, the
variable’s value should always be <span class="c000">../../..</span> regardless of what
it was in the HPL build script file.</p>
<!--TOC section id="sec2" Compiling-->
<h2 class="section" id="sec2">2  Compiling</h2><!--SEC END --><p>
The first step is to create a build script file that reflects
characteristics of your machine. This file is reused by all the
components of the HPC Challenge suite. The build script file should be
created in the <span class="c000">hpl</span> directory. This directory contains
instructions (the files <span class="c000">README</span> and <span class="c000">INSTALL</span>) on how
to create the build script file for your system. The
<span class="c000">hpl/setup</span> directory contains many examples of build script
files. A recommended approach is to copy one of them to the
<span class="c000">hpl</span> directory and if it doesn’t work then change it.</p><p>The build script file has a name that starts with <span class="c000">Make.</span>
prefix and usally ends with a suffix that identifies the target
system. For example, if the suffix chosen for the system is
<span class="c000">Unix</span>, the file should be named <span class="c000">Make.Unix</span>.</p><p>To build the benchmark executable (for the system named <span class="c000">Unix</span>)
type: <span class="c000">make arch=Unix</span>. This command should be run in the top
directory (not in the <span class="c000">hpl</span> directory). It will look in the
<span class="c000">hpl</span> directory for the build script file and use it to build
the benchmark executable.</p><p>The runtime behavior of the HPC Challenge source code may be
configured at compiled time by defining a few C preprocessor
symbols. They can be defined by adding appropriate options to
<span class="c000">CCNOOPT</span> and <span class="c000">CCFLAGS</span> make variables. The former
controls options for source code files that need to be compiled
without aggressive optimizations to ensure accurate generation of
system-specific parameters. The latter applies to the rest of the
files that need good compiler optimization for best performance. To
define a symbol <span class="c000">S</span>, the majority of compilers requires option
<span class="c000">-DS</span> to be used. Currently, the following options are
available in the HPC Challenge source code:
</p><ul class="itemize"><li class="li-itemize"><span class="c000">HPCC_FFT_235</span>: if this symbol is defined the FFTE
code (an FFT implementation) will use vector sizes and processor
counts that are not limited to powers of 2. Instead, the vector sizes
and processor counts to be used will be a product of powers of 2, 3,
and 5.</li><li class="li-itemize"><span class="c000">HPCC_FFTW_ESTIMATE</span>: if this symbol is defined it will
affect the way external FFTW library is called (it does not have any
effect if the FFTW library is not used). When defined, this symbol
will call the FFTW planning routine with <span class="c000">FFTW_ESTIMATE</span>
flag (instead of <span class="c000">FFTW_MEASURE</span>). This might result with worse
performance results but shorter execution time of the
benchmark. Defining this symbol may also positively affect the memory
fragmentation caused by the FFTW’s planning routine.</li><li class="li-itemize"><span class="c000">HPCC_MEMALLCTR</span>: if this symbol is defined a custom
memory allocator will be used to alleviate effects of memory
fragmentation and allow for larger data sets to be used which may
result in obtaining better performance.</li><li class="li-itemize"><span class="c000">HPL_USE_GETPROCESSTIMES</span>: if this symbol is defined
then Windows-specific <span class="c000">GetProcessTimes()</span> function will be used
to measure the elapsed CPU time.</li><li class="li-itemize"><span class="c000">USE_MULTIPLE_RECV</span>: if this symbol is defined then multiple non-blocking
receives will be posted simultaneously. By default only one non-blocking
receive is posted.</li><li class="li-itemize"><span class="c000">RA_SANDIA_NOPT</span>: if this symbol is defined the
HPC Challenge standard algorithm for Global RandomAccess will not be
used. Instead, an alternative implementation from Sandia
National Laboratory will be used. It routes messages in software
across virtual hyper-cube topology formed from MPI processes.</li><li class="li-itemize"><span class="c000">RA_SANDIA_OPT2</span>: if this symbol is defined the
HPC Challenge standard algorithm for Global RandomAccess will not be
used. Instead, instead an alternative implementation from Sandia
National Laboratory will be used. This implementation is optimized for
number of processors being powers of two. The optimizations
are sorting of data before sending and unrolling the data update
loop. If the number of process is not a power two then the code
is the same as the one performed with the <span class="c000">RA_SANDIA_NOPT</span> setting.</li><li class="li-itemize"><span class="c000">RA_TIME_BOUND_DISABLE</span>: if this symbol is defined then the
standard Global RandomAccess code will be used without time limits. This is
discouraged for most runs because the standard algorithm tends to be slow for
large array sizes due to a large overhead for short MPI messages.</li><li class="li-itemize"><span class="c000">USING_FFTW</span>: if this symbol is defined the standard
HPC Challenge FFT implemenation (called FFTE) will not be used.
Instead, FFTW library will be called. Defining the
<span class="c000">USING_FFTW</span> symbol is not sufficient: appropriate flags have
to be added in the make script so that FFTW headers files can be found
at compile time and the FFTW libraries at link time.</li></ul>
<!--TOC section id="sec3" Runtime Configuration-->
<h2 class="section" id="sec3">3  Runtime Configuration</h2><!--SEC END --><p>
The HPC Challenge is driven by a short input file named
<span class="c000">hpccinf.txt</span> that is almost the same as the input file for
HPL (customarily called <span class="c000">HPL.dat</span>). Refer to the directory
<span class="c000">hpl/www/tuning.html</span> for details about the input file for
HPL. A sample input file is included with the HPC Challenge
distribution.</p><p>The differences between HPL’s input file and HPC Challenge’s input
file can be summarized as follows:</p><ul class="itemize"><li class="li-itemize">
Lines 3 and 4 are ignored. The output is always appended to the
file named <span class="c000">hpccoutf.txt</span>.
</li><li class="li-itemize">There are additional lines (starting with line 33) that may (but
do not have to) be used to customize the HPC Challenge benchmark. They
are described below.
</li></ul><p>The additional lines in the HPC Challenge input file (compared to the
HPL input file) are:</p><ul class="itemize"><li class="li-itemize">
Lines 33 and 34 describe additional matrix sizes to be used for
running the PTRANS benchmark (one of the components of the
HPC Challenge benchmark).
</li><li class="li-itemize">Lines 35 and 36 describe additional blocking factors to be used
for running the PTRANS test.
</li></ul><p>Just for completeness, here is the list of lines of the HPC
Challenge’s input file and brief description of their meaning:
</p><ul class="itemize"><li class="li-itemize">
Line 1: ignored
</li><li class="li-itemize">Line 2: ignored
</li><li class="li-itemize">Line 3: ignored
</li><li class="li-itemize">Line 4: ignored
</li><li class="li-itemize">Line 5: number of matrix sizes for HPL (and PTRANS)
</li><li class="li-itemize">Line 6: matrix sizes for HPL (and PTRANS)
</li><li class="li-itemize">Line 7: number of blocking factors for HPL (and PTRANS)
</li><li class="li-itemize">Line 8: blocking factors for HPL (and PTRANS)
</li><li class="li-itemize">Line 9: type of process ordering for HPL
</li><li class="li-itemize">Line 10: number of process grids for HPL (and PTRANS)
</li><li class="li-itemize">Line 11: numbers of process rows of each process grid for HPL (and PTRANS)
</li><li class="li-itemize">Line 12: numbers of process columns of each process grid for HPL (and PTRANS)
</li><li class="li-itemize">Line 13: threshold value not to be exceeded by scaled residual for HPL (and PTRANS)
</li><li class="li-itemize">Line 14: number of panel factorization methods for HPL
</li><li class="li-itemize">Line 15: panel factorization methods for HPL
</li><li class="li-itemize">Line 16: number of recursive stopping criteria for HPL
</li><li class="li-itemize">Line 17: recursive stopping criteria for HPL
</li><li class="li-itemize">Line 18: number of recursion panel counts for HPL
</li><li class="li-itemize">Line 19: recursion panel counts for HPL
</li><li class="li-itemize">Line 20: number of recursive panel factorization methods for HPL
</li><li class="li-itemize">Line 21: recursive panel factorization methods for HPL
</li><li class="li-itemize">Line 22: number of broadcast methods for HPL
</li><li class="li-itemize">Line 23: broadcast methods for HPL
</li><li class="li-itemize">Line 24: number of look-ahead depths for HPL
</li><li class="li-itemize">Line 25: look-ahead depths for HPL
</li><li class="li-itemize">Line 26: swap methods for HPL
</li><li class="li-itemize">Line 27: swapping threshold for HPL
</li><li class="li-itemize">Line 28: form of L1 for HPL
</li><li class="li-itemize">Line 29: form of U for HPL
</li><li class="li-itemize">Line 30: value that specifies whether equilibration should be used by HPL
</li><li class="li-itemize">Line 31: memory alignment for HPL
</li><li class="li-itemize">Line 32: ignored
</li><li class="li-itemize">Line 33: number of additional problem sizes for PTRANS
</li><li class="li-itemize">Line 34: additional problem sizes for PTRANS
</li><li class="li-itemize">Line 35: number of additional blocking factors for PTRANS
</li><li class="li-itemize">Line 36: additional blocking factors for PTRANS
</li></ul>
<!--TOC section id="sec4" Running-->
<h2 class="section" id="sec4">4  Running</h2><!--SEC END --><p>
The exact way to run the HPC Challenge benchmark depends on the MPI
implementation and system details. An example command to run the
benchmark could like like this: <span class="c000">mpirun -np 4 hpcc</span>. The
meaning of the command’s components is as follows:
</p><ul class="itemize"><li class="li-itemize">
<span class="c000">mpirun</span> is the command that starts execution of an MPI
code. Depending on the system, it might also be <span class="c000">aprun</span>,
<span class="c000">mpiexec</span>, <span class="c000">mprun</span>, <span class="c000">poe</span>, or something
appropriate for your computer.</li><li class="li-itemize"><span class="c000">-np 4</span> is the argument that specifies that 4 MPI
processes should be started. The number of MPI processes should be
large enough to accomodate all the process grids specified in the
<span class="c000">hpccinf.txt</span> file.</li><li class="li-itemize"><span class="c000">hpcc</span> is the name of the HPC Challenge executable to
run.
</li></ul><p>After the run, a file called <span class="c000">hpccoutf.txt</span> is created. It
contains results of the benchmark. This file should be uploaded
through the web form at the HPC Challenge website.</p>
<!--TOC section id="sec5" Source Code Changes across Versions (ChangeLog)-->
<h2 class="section" id="sec5">5  Source Code Changes across Versions (ChangeLog)</h2><!--SEC END -->
<!--TOC subsection id="sec6" Version 1.5.0 (2016-03-18)-->
<h3 class="subsection" id="sec6">5.1  Version 1.5.0 (2016-03-18)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Fixed memory leak in STREAM code.
</li><li class="li-enumerate">Fixed bug in STREAM that resulted in minimum results reported as 0.
</li><li class="li-enumerate">Removed some of the compilation warnings.
</li></ol>
<!--TOC subsection id="sec7" Version 1.5.0beta (2015-07-23)-->
<h3 class="subsection" id="sec7">5.2  Version 1.5.0beta (2015-07-23)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Added new targets to the main make(1) file.
</li><li class="li-enumerate">Fixed bug introduced while updating to MPI STREAM 1.7 with spurious global communicator (reported by NEC).
</li><li class="li-enumerate">Added make(1) file for OpenMPI from MacPorts.
</li><li class="li-enumerate">Fixed bug introduced while updating to MPI STREAM 1.7 that caused some ranks to use NULL communicator.
</li><li class="li-enumerate">Fixed bug introduced while updating to MPI STREAM 1.7 that caused syntax errors.
</li></ol>
<!--TOC subsection id="sec8" Version 1.5.0alpha (2015-05-22)-->
<h3 class="subsection" id="sec8">5.3  Version 1.5.0alpha (2015-05-22)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Added global error accounting in STREAM.
</li><li class="li-enumerate">Updated checking to report from multiple MPI processes contributing to overall error.
</li><li class="li-enumerate">Added barrier to make sure all processes enter STREAM kernel tests at the same time.
</li><li class="li-enumerate">Updated naming conventions to match the original benchmark in STREAM.
</li><li class="li-enumerate">Changed scaling constant to prevent verification from overflowing in STREAM.
</li><li class="li-enumerate">Simplified MPI communicator code in STREAM.
</li><li class="li-enumerate">Substituted large constants for more descriptive compile time arithmetic in STREAM.
</li><li class="li-enumerate">Added the “restrict” keyword to the STREAM vector pointers for faster generated code.
</li><li class="li-enumerate">Updated STREAM code to the official STREAM MPI version 1.7.
</li><li class="li-enumerate">Removed infinite loop due to default compiler optimization in DLAMCH and SLAMCH.
</li><li class="li-enumerate">Added compiler flags to allow compiling with a C++ compiler.
</li></ol>
<!--TOC subsection id="sec9" Version 1.4.3 (2013-08-26)-->
<h3 class="subsection" id="sec9">5.4  Version 1.4.3 (2013-08-26)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Increased the size of scratch vector for local FFT tests that was
missed in the previous version (reported by SGI).
</li><li class="li-enumerate">Added Makefile for Blue Gene/P contributed by Vasil Tsanov.
</li></ol>
<!--TOC subsection id="sec10" Version 1.4.2 (2012-10-12)-->
<h3 class="subsection" id="sec10">5.5  Version 1.4.2 (2012-10-12)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Increased sizes of scratch vectors for local FFT tests to account for
runs on systems with large main memory (reported by IBM, SGI and Intel).
</li><li class="li-enumerate">Reduced vector size for local FFT tests due to larger scratch space needed.
</li><li class="li-enumerate">Added a type cast to prevent overflow of a 32-bit integer vector
size in FFT data generation routine (reported by IBM).
</li><li class="li-enumerate">Fixed variable types to handle array sizes that overflow 32-bit
integers in RandomAccess (reported by IBM and SGI).
</li><li class="li-enumerate">Changed time-bound code to be used by default in Global RandomAccess and
allowed for it to be switched off with a compile time flag if necessary.
</li><li class="li-enumerate">Code cleanup to allow compilation without warnings of RandomAccess test.
</li><li class="li-enumerate">Changed communication code in PTRANS to avoid large message sizes that
caused problems in some MPI implementations.
</li><li class="li-enumerate">Updated documentation in README.txt and README.html files.
</li></ol>
<!--TOC subsection id="sec11" Version 1.4.1 (2010-06-01)-->
<h3 class="subsection" id="sec11">5.6  Version 1.4.1 (2010-06-01)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Added optimized variants of RandomAccess that use Linear Congruential Generator for random number generation.
</li><li class="li-enumerate">Made corrections to comments that provide definition of the RandomAccess test.
</li><li class="li-enumerate">Removed initialization of the main array from the timed section of optimized versions of RandomAccess.
</li><li class="li-enumerate">Fixed the length of the vector used to compute error when using MPI implementation from FFTW.
</li><li class="li-enumerate">Added global reduction to error calculation in MPI FFT to achieve more accurate error estimate.
</li><li class="li-enumerate">Updated documentation in README.
</li></ol>
<!--TOC subsection id="sec12" Version 1.4.0 (2010-03-26)-->
<h3 class="subsection" id="sec12">5.7  Version 1.4.0 (2010-03-26)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Added new variant of RandomAccess that uses Linear Congruential Generator for random number generation.
</li><li class="li-enumerate">Rearranged the order of benchmarks so that HPL component runs last and may be aborted
if the performance of other components was not satisfactory. RandomAccess is now first to assist in tuning
the code.
</li><li class="li-enumerate">Added global initialization and finalization routine that allows to properly initialize
and finalize external software and hardware components without changing the rest of the HPCC testing harness.
</li><li class="li-enumerate">Lack of <span class="c000">hpccinf.txt</span> is no longer reported as error but as a warning.
</li></ol>
<!--TOC subsection id="sec13" Version 1.3.2 (2009-03-24)-->
<h3 class="subsection" id="sec13">5.8  Version 1.3.2 (2009-03-24)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Fixed memory leaks in G-RandomAccess driver routine.
</li><li class="li-enumerate">Made the check for 32-bit vector sizes in G-FFT optional. MKL allows for 64-bit vector sizes in its FFTW wrapper.
</li><li class="li-enumerate">Fixed memory bug in single-process FFT.
</li><li class="li-enumerate">Update documentation (README).
</li></ol>
<!--TOC subsection id="sec14" Version 1.3.1 (2008-12-09)-->
<h3 class="subsection" id="sec14">5.9  Version 1.3.1 (2008-12-09)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Fixed a dead-lock problem in FFT component due to use of wrong communicator.
</li><li class="li-enumerate">Fixed the 32-bit random number generator in PTRANS that was using 64-bit
routines from HPL.
</li></ol>
<!--TOC subsection id="sec15" Version 1.3.0 (2008-11-13)-->
<h3 class="subsection" id="sec15">5.10  Version 1.3.0 (2008-11-13)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Updated HPL component to use HPL 2.0 source code
<ol class="enumerate" type=a><li class="li-enumerate">
Replaced 32-bit Pseudo Random Number Generator (PRNG) with a 64-bit one.
</li><li class="li-enumerate">Removed 3 numerical checks of the solution residual with a single one.
</li><li class="li-enumerate">Added support for 64-bit systems with large memory sizes (before they would
overflow during index calculations 32-bit integers.)
</li></ol>
</li><li class="li-enumerate">Introduced a limit on FFT vector size so they fit in a 32-bit integer (only
applicable when using FFTW version 2.)
</li></ol>
<!--TOC subsection id="sec16" Version 1.2.0 (2007-06-25)-->
<h3 class="subsection" id="sec16">5.11  Version 1.2.0 (2007-06-25)</h3><!--SEC END --><ol class="enumerate" type=1><li class="li-enumerate">
Changes in the FFT component:
<ol class="enumerate" type=a><li class="li-enumerate">
Added flexibility in choosing vector sizes and processor counts:
now the code can do powers of 2, 3, and 5 both sequentially and in parallel
tests.
</li><li class="li-enumerate">FFTW can now run with ESTIMATE (not just MEASURE) flag: it might produce
worse performance results but often reduces time to run the test and cuases
less memory fragmentation.
</li></ol>
</li><li class="li-enumerate">Changes in the DGEMM component:
<ol class="enumerate" type=a><li class="li-enumerate">
Added more comprehensive checking of the numerical properties of the
test’s results.
</li></ol>
</li><li class="li-enumerate">Changes in the RandomAccess component:
<ol class="enumerate" type=a><li class="li-enumerate">
Removed time-bound functionality: only runs that perform complete
computation are now possible.
</li><li class="li-enumerate">Made the timing more accurate: main array initialization is not counted
towards performance timing.
</li><li class="li-enumerate">Cleaned up the code: some non-portable C language constructs have been
removed.
</li><li class="li-enumerate">Added new algorithms: new algorithms from Sandia based on hypercube
network topology can now be chosen at compile time which results on much
better performance results on many types of parallel systems.
</li><li class="li-enumerate">Fixed potential resource leaks by adding function calls rquired by the MPI
standard.
</li></ol>
</li><li class="li-enumerate">Changes in the HPL component:
<ol class="enumerate" type=a><li class="li-enumerate">
Cleaned up reporting of numerics: more accurate printing of scaled
residual formula.
</li></ol>
</li><li class="li-enumerate">Changes in the PTRANS component:
<ol class="enumerate" type=a><li class="li-enumerate">
Added randomization of virtual process grids to measure bandwidth of the
network more accurately.
</li></ol>
</li><li class="li-enumerate">Miscellaneous changes:
<ol class="enumerate" type=a><li class="li-enumerate">
Added better support for Windows-based clusters by taking advantage of
Win32 API.
</li><li class="li-enumerate">Added custom memory allocator to deal with memory fragmentation on some
systems.
</li><li class="li-enumerate">Added better reporting of configuration options in the output file.
</li></ol>
</li></ol>
<!--TOC subsection id="sec17" Version 1.0.0 (2005-06-11)-->
<h3 class="subsection" id="sec17">5.12  Version 1.0.0 (2005-06-11)</h3><!--SEC END -->
<!--TOC subsection id="sec18" Version 0.8beta (2004-10-19)-->
<h3 class="subsection" id="sec18">5.13  Version 0.8beta (2004-10-19)</h3><!--SEC END -->
<!--TOC subsection id="sec19" Version 0.8alpha (2004-10-15)-->
<h3 class="subsection" id="sec19">5.14  Version 0.8alpha (2004-10-15)</h3><!--SEC END -->
<!--TOC subsection id="sec20" Version 0.6beta (2004-08-21)-->
<h3 class="subsection" id="sec20">5.15  Version 0.6beta (2004-08-21)</h3><!--SEC END -->
<!--TOC subsection id="sec21" Version 0.6alpha (2004-05-31)-->
<h3 class="subsection" id="sec21">5.16  Version 0.6alpha (2004-05-31)</h3><!--SEC END -->
<!--TOC subsection id="sec22" Version 0.5beta (2003-12-01)-->
<h3 class="subsection" id="sec22">5.17  Version 0.5beta (2003-12-01)</h3><!--SEC END -->
<!--TOC subsection id="sec23" Version 0.4alpha (2003-11-13)-->
<h3 class="subsection" id="sec23">5.18  Version 0.4alpha (2003-11-13)</h3><!--SEC END -->
<!--TOC subsection id="sec24" Version 0.3alpha (2004-11-05)-->
<h3 class="subsection" id="sec24">5.19  Version 0.3alpha (2004-11-05)</h3><!--SEC END --><!--BEGIN NOTES document-->
<hr class="footnoterule"><dl class="thefootnotes"><dt class="dt-thefootnotes">
<a id="note1" href="#text1">*</a></dt><dd class="dd-thefootnotes"><div class="footnotetext">University of Tennessee Knoxville, Innovative
Computing Laboratory</div></dd></dl>
<!--END NOTES-->
<!--CUT END -->
<!--HTMLFOOT-->
<!--ENDHTML-->
<!--FOOTER-->
<hr style="height:2"><blockquote class="quote"><em>This document was translated from L<sup>A</sup>T<sub>E</sub>X by
</em><a href="http://hevea.inria.fr/index.html"><em>H<span style="font-size:small"><sup>E</sup></span>V<span style="font-size:small"><sup>E</sup></span>A</em></a><em>.</em></blockquote></body>
</html>
|
