| 1 |
%z Article |
| 2 |
%K Staelin02b |
| 3 |
%A Carl Staelin |
| 4 |
%T lmbench3: Measuring scalability |
| 5 |
%D November 2002 |
| 6 |
%I Hewlett-Packard Laboratories |
| 7 |
%C Palo Alto, CA |
| 8 |
|
| 9 |
%z Article |
| 10 |
%K Staelin02c |
| 11 |
%A Carl Staelin |
| 12 |
%T Utilizing intra-processor parallelism |
| 13 |
%D December 2002 |
| 14 |
%I Hewlett-Packard Laboratories |
| 15 |
%C Palo Alto, CA |
| 16 |
|
| 17 |
%z Article |
| 18 |
%K Whaley98 |
| 19 |
%A R. Clint Whaley |
| 20 |
%A Jack Dongarra |
| 21 |
%T Automatically tuned linear algebra software |
| 22 |
%C Proceedings of the 1998 ACM/IEEE SC98 Conference |
| 23 |
%D 1998 |
| 24 |
%O http://sourceforge.net/projects/math-atlas |
| 25 |
|
| 26 |
%z Article |
| 27 |
%K Staelin98 |
| 28 |
%A Carl Staelin |
| 29 |
%A Larry McVoy |
| 30 |
%T mhz: Anatomy of a microbenchmark |
| 31 |
%C Proceedings USENIX Annual Technical Conference |
| 32 |
%c New Orleans, LA |
| 33 |
%D June 1998 |
| 34 |
%P 155-166 |
| 35 |
|
| 36 |
%z Article |
| 37 |
%K McVoy96 |
| 38 |
%A Larry McVoy |
| 39 |
%A Carl Staelin |
| 40 |
%T lmbench: Portable tools for performance analysis |
| 41 |
%C Proceedings USENIX Winter Conference |
| 42 |
%c San Diego, CA |
| 43 |
%D January 1996 |
| 44 |
%P 279-284 |
| 45 |
|
| 46 |
%a Thesis |
| 47 |
%K Prestor01 |
| 48 |
%A Uros Prestor |
| 49 |
%T Evaluating the memory performance of a ccNUMA system |
| 50 |
%R Masters Thesis |
| 51 |
%I School of Computing, University of Utah |
| 52 |
%c Salt Lake City, Utah |
| 53 |
%D May 2001 |
| 54 |
%O http://www.cs.utah.edu/~uros/thesis/thesis.pdf |
| 55 |
|
| 56 |
%z Article |
| 57 |
%K Saavedra95 |
| 58 |
%A R.H. Saavedra |
| 59 |
%A A.J. Smith |
| 60 |
%T Measuring cache and TLB performance and their effect on benchmark runtimes |
| 61 |
%J IEEE Transactions on Computers |
| 62 |
%V 44 |
| 63 |
%N 10 |
| 64 |
%D October 1995 |
| 65 |
%P 1223-1235 |
| 66 |
|
| 67 |
%z Article |
| 68 |
%K Wolman89 |
| 69 |
%A Barry L. Wolman |
| 70 |
%A Thomas M. Olson |
| 71 |
%T IOBENCH: a system independent IO benchmark |
| 72 |
%J Computer Architecture News |
| 73 |
%V 17 |
| 74 |
%N 5 |
| 75 |
%D September 1989 |
| 76 |
%P 55-70 |
| 77 |
%x IOBENCH is an operating system and processor independent synthetic |
| 78 |
%x input/output (IO) benchmark designed to put a configurable IO and |
| 79 |
%x processor (CP) load on the system under test. This paper discusses |
| 80 |
%x the UNIX versions. |
| 81 |
%k IOBENCH, synthetic I/O benchmark, UNIX workload |
| 82 |
%s vinton%[email protected] (Fri Sep 20 12:55:58 PDT 1991) |
| 83 |
|
| 84 |
%z Book |
| 85 |
%K Hennessy96 |
| 86 |
%A John L. Hennessy |
| 87 |
%A David A. Patterson |
| 88 |
%T Computer Architecture A Quantitative Approach, 2nd Edition |
| 89 |
%I Morgan Kaufman |
| 90 |
%D 1996 |
| 91 |
|
| 92 |
%z Article |
| 93 |
%K Chen94a |
| 94 |
%A P. M. Chen |
| 95 |
%A D. A. Patterson |
| 96 |
%T A new approach to I/O performance evaluation \- self-scaling I/O benchmarks, predicted I/O performance |
| 97 |
%D November 1994 |
| 98 |
%J Transactions on Computer Systems |
| 99 |
%V 12 |
| 100 |
%N 4 |
| 101 |
%P 308-339 |
| 102 |
%x Current I/O benchmarks suffer from several chronic problems: they |
| 103 |
%x quickly become obsolete; they do not stress the I/O system; and they |
| 104 |
%x do not help much in undelsi;anding I/O system performance. We |
| 105 |
%x propose a new approach to I/O performance analysis. First, we |
| 106 |
%x propose a self-scaling benchmark that dynamically adjusts aspects of |
| 107 |
%x its workload according to the performance characteristic of the |
| 108 |
%x system being measured. By doing so, the benchmark automatically |
| 109 |
%x scales across current and future systems. The evaluation aids in |
| 110 |
%x understanding system performance by reporting how performance varies |
| 111 |
%x according to each of five workload parameters. Second, we propose |
| 112 |
%x predicted performance, a technique for using the results from the |
| 113 |
%x self-scaling evaluation to estimate quickly the performance for |
| 114 |
%x workloads that have not been measured. We show that this technique |
| 115 |
%x yields reasonably accurate performance estimates and argue that this |
| 116 |
%x method gives a far more accurate comparative performance evaluation |
| 117 |
%x than traditional single-point benchmarks. We apply our new |
| 118 |
%x evaluation technique by measuring a SPARCstation 1+ with one SCSI |
| 119 |
%x disk, an HP 730 with one SCSI-II disk, a DECstation 5000/200 running |
| 120 |
%x the Sprite LFS operating system with a three-disk disk array, a |
| 121 |
%x Convex C240 minisupercomputer with a four-disk disk array, and a |
| 122 |
%x Solbourne 5E/905 fileserver with a two-disk disk array. |
| 123 |
%s [email protected] (Mon Mar 13 10:57:38 1995) |
| 124 |
%s wilkes%[email protected] (Sun Mar 19 12:38:01 PST 1995) |
| 125 |
%s wilkes%[email protected] (Sun Mar 19 12:38:53 PST 1995) |
| 126 |
|
| 127 |
%z InProceedings |
| 128 |
%K Ousterhout90 |
| 129 |
%s wilkes%[email protected] (Fri Jun 29 20:46:08 PDT 1990) |
| 130 |
%A John K. Ousterhout |
| 131 |
%T Why aren't operating systems getting faster as fast as hardware? |
| 132 |
%C Proceedings USENIX Summer Conference |
| 133 |
%c Anaheim, CA |
| 134 |
%D June 1990 |
| 135 |
%P 247-256 |
| 136 |
%x This paper evaluates several hardware pplatforms and operating systems using |
| 137 |
%x a set of benchmarks that stress kernel entry/exit, file systems, and |
| 138 |
%x other things related to operating systems. The overall conclusion is that |
| 139 |
%x operating system performance is not improving at the same rate as the base speed of the |
| 140 |
%x underlying hardware. The most obvious ways to remedy this situation |
| 141 |
%x are to improve memory bandwidth and reduce operating systems' |
| 142 |
%x tendency to wait for disk operations to complete. |
| 143 |
%o Typical performance of 10-20 MIPS cpus is only 0.4 times what |
| 144 |
%o their raw hardware performance would suggest. HP-UX is |
| 145 |
%o particularly bad on the HP 9000/835, at about 0.2x. (Although |
| 146 |
%o this measurement discounted a highly-tuned getpid call.) |
| 147 |
%k OS performance, RISC machines, HP9000 Series 835 system calls |
| 148 |
|
| 149 |
%z InProceedings |
| 150 |
%K McVoy91 |
| 151 |
%A L. W. McVoy |
| 152 |
%A S. R. Kleiman |
| 153 |
%T Extent-like Performance from a Unix File System |
| 154 |
%C Proceedings USENIX Winter Conference |
| 155 |
%c Dallas, TX |
| 156 |
%D January 1991 |
| 157 |
%P 33-43 |
| 158 |
|
| 159 |
%z Article |
| 160 |
%K Chen93d |
| 161 |
%A Peter M. Chen |
| 162 |
%A David Patterson |
| 163 |
%T Storage performance \- metrics and benchmarks |
| 164 |
%J Proceedings of the IEEE |
| 165 |
%V 81 |
| 166 |
%N 8 |
| 167 |
%D August 1993 |
| 168 |
%P 1151-1165 |
| 169 |
%x Discusses metrics and benchmarks used in storage performance evaluation. |
| 170 |
%x Describes, reviews, and runs popular I/O benchmarks on three systems. Also |
| 171 |
%x describes two new approaches to storage benchmarks: LADDIS and a Self-Scaling |
| 172 |
%x benchmark with predicted performance. |
| 173 |
%k I/O, storage, benchmark, workload, self-scaling benchmark, |
| 174 |
%k predicted performance, disk, performance evaluation |
| 175 |
%s staelin%[email protected] (Wed Sep 27 16:21:11 PDT 1995) |
| 176 |
|
| 177 |
%z Article |
| 178 |
%K Park90a |
| 179 |
%A Arvin Park |
| 180 |
%A J. C. Becker |
| 181 |
%T IOStone: a synthetic file system benchmark |
| 182 |
%J Computer Architecture News |
| 183 |
%V 18 |
| 184 |
%N 2 |
| 185 |
%D June 1990 |
| 186 |
%P 45-52 |
| 187 |
%o this benchmark is useless for all modern systems; it fits |
| 188 |
%o completely inside the file system buffer cache. Soon it may even |
| 189 |
%o fit inside the processor cache! |
| 190 |
%k IOStone, I/O, benchmarks |
| 191 |
%s staelin%[email protected] (Wed Sep 27 16:37:26 PDT 1995) |
| 192 |
|
| 193 |
%z Article |
| 194 |
%K Fenwick95 |
| 195 |
%A David M. Fenwick |
| 196 |
%A Denis J. Foley |
| 197 |
%A William B. Gist |
| 198 |
%A Stephen R. VanDoren |
| 199 |
%A Danial Wissell |
| 200 |
%T The AlphaServer 8000 series: high-end server platform development |
| 201 |
%J Digital Technical Journal |
| 202 |
%V 7 |
| 203 |
%N 1 |
| 204 |
%D August 1995 |
| 205 |
%P 43-65 |
| 206 |
%x The AlphaServer 8400 and the AlphaServer 8200 are Digital's newest high-end |
| 207 |
%x server products. Both servers are based on the 300Mhz Alpha 21164 |
| 208 |
%x microprocessor and on the AlphaServer 8000-series platform architecture. |
| 209 |
%x The AlphaServer 8000 platform development team set aggressive system data |
| 210 |
%x bandwidth and memory read latency targets in order to achieve high-performance |
| 211 |
%x goals. The low-latency criterion was factored into design decisions made at |
| 212 |
%x each of the seven layers of platform development. The combination of |
| 213 |
%x industry-leading microprocessor technology and a system platform focused |
| 214 |
%x on low latency has resulted in a 12-processor server implementation --- |
| 215 |
%x the AlphaServer 8400 --- capable of supercomputer levels of performance. |
| 216 |
%k DEC Alpha server, performance, memory latency |
| 217 |
%s staelin%[email protected] (Wed Sep 27 17:27:23 PDT 1995) |
| 218 |
|
| 219 |
%z Book |
| 220 |
%K Toshiba94 |
| 221 |
%A Toshiba |
| 222 |
%T DRAM Components and Modules |
| 223 |
%I Toshiba America Electronic Components, Inc. |
| 224 |
%P A59-A77,C37-C42 |
| 225 |
%D 1994 |
| 226 |
|
| 227 |
%z Article |
| 228 |
%K McCalpin95 |
| 229 |
%A John D. McCalpin |
| 230 |
%T Memory bandwidth and machine balance in current high performance computers |
| 231 |
%J IEEE Technical Committee on Computer Architecture newsletter |
| 232 |
%D December 1995 |
| 233 |
|
| 234 |
%z Article |
| 235 |
%K FSF89 |
| 236 |
%A Richard Stallman |
| 237 |
%Q Free Software Foundation |
| 238 |
%T General Public License |
| 239 |
%D 1989 |
| 240 |
%O Included with \*[lmbench] |
| 241 |
|
| 242 |
%z Article |
| 243 |
%K Min01 |
| 244 |
%A Rui Min |
| 245 |
%A Yiming Hu |
| 246 |
%T Improving performance of large physically indexed caches by decoupling memory addresses from cache addresses |
| 247 |
%J IEEE Transactions on Computers |
| 248 |
%V 50 |
| 249 |
%N 11 |
| 250 |
%D November 2001 |
| 251 |
%P 1191-1201 |