source: Sophya/trunk/SophyaLib/Manual/perfmachine.txt@ 3229

Last change on this file since 3229 was 3187, checked in by ansari, 19 years ago

Ajout mesures avec zthr sync - Reza 20/02/2007

File size: 13.1 KB
RevLine 
[3158]1-------------------------------------------------------------------------------------
2 Comparaison performance de differentes machines en compilation / execution (calcul)
3 -----------------------
4 Mesures effectuees en Janvier 2007 , R. Ansari / C. Magneville
5-------------------------------------------------------------------------------------
6
7(a) eros3 : Bipro-bicoeur Xeon@2.4 GHz Linux (xeon-lx-2.4GHz) , gcc 3.2
8(b) ccali : Bipro-bicoeur Xeon@2.8 GHz Linux (xeon-lx-2.8GHz) , icc 8.0 ou 9.0
[3179]9(c) sgsda: AMD Bipro AMD opteron 248 @ 2.2 GHz (amd-lx-
[3187]10(cc) grid-saclay A: MD opteron 275 Bipro-bicoeur @ 2.2 GHz (amd275-lx)
[3158]11
12(d) asc: bipro alpha (@ ~1 GHz) server DS20 OSF (osf1) , cxx 6.5 (osf-asc)
13(nouveau asc 420 MFLOPS, moins puissante que l'ancien asc 800 MFLOPS)
14(e) xp1000-dapnia: alpha xp1000 @ ~ 600 MHz ? OSF1 , cxx ? (osf-xp1000)
15(f) superosf-dapnia: multi-proc alphaServer ES80 6 procs EV7 @ 1 GHz (super-osf)
16
17(g) ccsvx01: XServe G5 bipro @~1GHz (Darwin/OSX) (G5-osx-1GHz) , gcc 3.3
18(h) PowerBook-Reza : Apple G4 @ 1.25 GHz (G4-osx-1.25GHz) , gcc 3.3
19(i) MacBook-Reza: Apple/ Core double-coeur Intel @ 1.83 GHz (core-osx-1.83GHz) gcc 4
20(j) MacPro-Grosdidier : Apple / Xeon 2 double-coeur @ 3 GHz gcc 4.0.1 , compil SOPHYA -O2 -g
21
[3187]22(p) IBM-AIX regatta , xlC , IBM eServer pSeries 655 , 8 proc power4 @ 1.1 GHz
[3158]23
24
25NOTES :
26- Sur les machines Xeon, il y a une interaction entre process / threads par rapport a
27l'occupation des CPU's. On perd un facteur 3 en performance multi-threads/multi-taches.
28La machine MacPro avec OSX se debrouille quand meme mieux.
29- Effet du systeme ou carte mere ???
30
[3187]31
32Donnees SPECint2000 (3) / SPECfp2000 (2) (http//www.spec.org)
33(1) MFLOPS -> cpupower 2
34----------------------------------------------------------------------
35 MFLOPS(1) SPECfp SPECint
36----------------------------------------------------------------------
37(b)xeon-lx-2.8GHz 900 1400 1400
38(c)amd-lx 690 1600 1300
39(cc)amd2-lx 675 1600 1300
40
41(d)osf-xp1000 648 500 400
42(e)superosf 842 1100 700
43
44(i)core-osx-1.83GHz 855 1400 1500
45(j)xeon-osx 2600 2900 -
46
47(p)ibm-aix 700 1050 700
48----------------------------------------------------------------------
49
50
[3158]51A/ Compilation tout SOPHYA
52----------------------------
53csh> time make all (1)
54ou
55csh> time make -j 2 all (2)
56 Temps CPU
57 Indice de performance 100*(1000/TCPU)
58 Temps elapsed (vrai)
59 Temps vrai / TCPU
60
61
62----------------------------------------------------------------------
63 CPU(s) IndPerf TElapsed , TCPU/Elapsed %
64----------------------------------------------------------------------
65(a)xeon-lx-2.4GHz (2) 615 s 162 410 s 150%
66 avec -O3 -g (2) 1300 s 77 760 s 172%
67(b)xeon-lx-2.8GHz (2) 755 s 132 540 s 140%
68(c)amd-lx
69
70(d)osf-xp1000 (1) 533 s 187 660 s 80%
71(e)superosf (1) 895 s 112 910 s 98%
72(f)osf-asc (1) 1920 s 52 2340 s 83% (??)
73
74(f)G4-osx-1.25GHz (1) 660 s 151 710 s 93%
75(h)G5-osx-1GHz (2) 453 s 221 250 s 182%
[3159]76 -tune=G5 1100 s 90
[3158]77(i)core-osx-1.83GHz (2) 209 s 478 116 s 180%
[3159]78 -O2 (1) 367 s 272 381 96%
[3158]79(j)xeon-osx
80
81(p)ibm-aix
82----------------------------------------------------------------------
83
84Taille shared libs :
85(a)
86(f) = (e) = 57 MO
87(h) 80 MO
88(i) 83 MO
89
[3187]90B/ Calcul brut (Tableaux de SOPHYA) avec / sans threads
91--------------------------------------------------------
[3158]92
93B.1/ arr = c1*a1+c2*a2
94(1) time cpupower 2 # compile avec -O3
95(2) time zthr arr 1 1000 1 thread
96(3) time zthr arr 2 1000 2 thread
97(4) time zthr arr 4 1000 4 thread
98(5) time zthr arr 6 1000 6 thread
99(6) time zthr arr 8 1000 8 thread
100
101-----------------------------------------------------------------------------------
102 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
103-----------------------------------------------------------------------------------
104(a)xeon-lx-2.4GHz 1167 5.15/5.2/99% 11.4/5.8/196%
105 (4)36.6/9.28/394%
106 -O3 -g 4.9/5./99%
107(b)xeon-lx-2.8GHz 920 2.3/2.3/100% 6.2/3.1/198%
108 (4)26/6.6/396%
109(c)amd-lx 690 3.6/3.6/99% 6.8/4/171%
110 (4)13.5/7/193%
111 (5)20.3/10.23/198%
[3187]112(cc)amd2-lx 675 2/2/99% 4.15/2.1/197%
113 (4)8.25/4.15/198%
114 (5)13.6/4.6/292%
115 (6)19.8/6.5/300%
[3158]116
117(d)osf-xp1000 648 5.1/5.3/96.6% 11.4/11.4/99%
118 (4)25.2/25.5/99%
119(e)superosf 842 2.87/2.88/99.6% 6.25/4.1/153%
120 (4)11.6/3.06/379%
121(f)osf-asc 420 6.3s/6.5s/99% 16.9/8.8/192%
122 (4)29.9/15.7/191%
123
124(f)G4-osx-1.25GHz 333 44s/48s/91% 86.7/99.8/92%
125(h)G5-osx-1GHz 1151 20s/20s/99% 40s/23s/170%
126 (4) 80.8/45/180%
[3159]127 -tune=G5 3.35/3.8/88% 7.1/3.6/196%
128 (4) 14/7.5/187%
[3158]129(i)core-osx-1.83GHz 855 11.5/11.5/100% 23/11.6/192%
130 (4) 46/23/199%
[3159]131 -O2 3.85/3.89/99% 7.7/3.9/198%
132 (4) 15.4/7.77/198%
133
[3158]134(j)xeon-osx 2600 2.5/2.5/100% 5.1/2.6/199%
135 (4) 11.5/3.2/362%
136 (5) 17.4/4.77/365%
137
138(p)ibm-aix-regatta 700 6.8/6.9/98% 13.1/6.75/195%
139 (4) 26.3/11.7/225%
140
[3179]141(q)sgi-magique 460 60/60/99%
[3158]142 -----------------------------------------------------------------------------------
143
144
145B.2/ Multiplication de matrices mtx = mtx1 * mtx2
146
147(1) time cpupower 2
148(2) time zthr mtx 1 1000 1 thread
149(3) time zthr mtx 2 1000 2 thread
150(4) time zthr mtx 4 1000 4 thread
151(5) time zthr mtx 6 1000 6 thread
152(6) time zthr mtx 8 1000 8 thread
153
154-----------------------------------------------------------------------------------
155 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
156-----------------------------------------------------------------------------------
157(a)xeon-lx-2.4GHz 1167 6.5/6.5/100% 17.4/8.8/198%
158 (4) 80.5/20.3/397%
159 (5) 114.5/29.6/387%
160 (6) 160/40.3/387%
161
162(b)xeon-lx-2.8GHz 920 3.4/3.4/100% 12/6.1/199%
163 (4) 55.8/14/400%
164(c)amd-lx 690 6.98/6.98/100% 14.1/8.15/173%
165 (4) 27.7/14.23/194%
166 (5) 41.4/21.07/196%
167 (6) 55.4/27.9/198.7%
[3187]168(cc)amd2-lx 675 4.1/4.1/100% 9.55/4.8/198%
169 (4) 20/10.27/195%
170 (5) 32.8/11.16/294%
171 (6) 42.75/13.8/309%
[3158]172
[3187]173
[3158]174(d)osf-xp1000 648 13/14.1/92% 27.1/27.4/99%
175 (4) 54/54.7/99.6%
176 (5) 80.6/81/99.6%
177 (6) 107.8/108.3/99.5%
178(e)superosf 842 6.1/7.24/84% 12.35/6.29/196%
179 (4) 24.3/6.31/385%
180 (5) 36.5/10.9/335%
181 (6) 50.1/18.15/276%
182(f)osf-asc 420 13.5s/13.7s/98% 32/16.5/194%
183 (4) 67.5/34.4/196%
184
185(f)G4-osx-1.25GHz 333
186(h)G5-osx-1GHz 1151 23/23.7/97% 46.5/27.5/170%
187 (4) 93.4/49.4/189%
[3159]188 -tune=G5 5.7/5.8/98% 13.3/6.8/197%
189 (4) 26.8/13.56/197%
190 (6) 53.8/27.25/197%
[3158]191(i)core-osx-1.83GHz 855 12.6/12.7/100% 25.8/13.4/194%
192 (4) 51.6/26/199%
[3159]193 -O2 4.25/4.5/94% 10.6/5.36/198%
194 (4) 20.87/10.68/198%
195 -O2 2 jobs // 2 x 5/5.4/92%
[3158]196(j)xeon-osx 2600 2.8/2.8/99% 9.3/4.66/199%
197 (4) 31.4/8.6/364%
198 (5) 47.1/12.96/364%
199 (6) 62.8/17.38/362%
200
201(p)ibm-aix-regatta 700 9.5/9.7/98% 18.3/16.0/114%
202 (4) 38.3/24.7/155%
[3179]203
204(q)sgi-magique 460 49/49/99% 101/56/181%
205
[3158]206 -----------------------------------------------------------------------------------
207
[3187]208C/ Calcul fft (FFTW , FFTPack )
209-------------------------------
[3158]210
211(1) time cpupower 2
212(2) time tfft 2000000 W d 0 0 (avec FFTW)
213(3) time tfft 2000000 P d 0 0 (avec FFTPack_Sophya)
214
215IndPerf=1000/TCPU
216
217
218-----------------------------------------------------------------------------------
219 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
220-----------------------------------------------------------------------------------
221(a)xeon-lx-2.4GHz 1167 5.5/5.6/97% 180 7.4/7.4/100% 135
222 -O3 -g 6.8/7.1/96% 147
223(b)xeon-lx-2.8GHz 920
224(c)amd-lx 690 4.2/4.2/100% 238
[3159]225 ~2x 4.7/4.7/99%
[3187]226(cc)amd2-lx 675 2.8/2.8/99% 3.56/3.58/99%
[3158]227
228(d)osf-xp1000 648 9.9/10.2/97% 101 9.3/9.46/98.5% 107
229(e)superosf 842 6./6.22/96.7% 166 5.1/5.18/97.4% 196
230(f)osf-asc 420 13.3/13.9/95.7% 75 12.2/17.6/70% 82
231
232(f)G4-osx-1.25GHz 333 15.2/15.9/96% 66 23.8/34.1/70% 42
233(h)G5-osx-1GHz 1151 8.5/8.7/97.5% 120 11.5/11.6/99% 87
[3159]234 -tune=G5 5/5.1/98% 200 5.5/5.6/97% 180
[3158]235(i)core-osx-1.83GHz 855 4.6/4.75/97% 217 7/7.08/99% 142
[3159]236 -O2 3.2/3.2/99% 312 3.6/3.6/99% 277
237 -O2 2 jobs // 2 x 3.9/4.3/90% 2x 4.7/5/91% 200
[3158]238(j)xeon-osx 2600 2.6/2.6/98% 384
[3159]239 2 jobs // ~2x 3.6/4.1/87% 250
240 4 jobs //
[3158]241
242(p)ibm-aix-regatta 700 6.25/18.9/33% 160 5.25/15.7/33% 190
243
[3179]244(q)sgi-magique 460 22/22/98% 42 24.5/25/99% 40
[3158]245 -----------------------------------------------------------------------------------
[3187]246
247
248D/ Efficacite de gestion de lock (mutex) avec les threads
249-----------------------------------------------------------
250(32 threads - operant sur 2000 vecteurs ~ 64000 lock/unlock/wait/broadcast)
251
252csh> time zthr syncp 32 2000 4
253(1) time cpupower 2
254
255----------------------------------------------------------------------
256 (1)MFLOPS CPU(s) IndPerf TCPU/Elapsed %
257----------------------------------------------------------------------
258(a)xeon-lx-2.4GHz 1167 17.8 178%
259 avec -O3 -g
260(b)xeon-lx-2.8GHz (2)
261(c)amd-lx 690 0.4 31%
262
263(d)osf-xp1000 (1)
264(e)superosf (1)
265(f)osf-asc (1) 420 5.4 100%
266
267(f)G4-osx-1.25GHz (1) 333 64 96%
268(h)G5-osx-1GHz (2) 1150 2.6 130%
269 -tune=G5
270(i)core-osx-1.83GHz
271 -O2
272(j)xeon-osx 2600 2.55 143%
273
274(p)ibm-aix 700 4.7 111%
275----------------------------------------------------------------------
Note: See TracBrowser for help on using the repository browser.