source: Sophya/trunk/SophyaLib/Manual/perfmachine.txt@ 3244

Last change on this file since 3244 was 3244, checked in by ansari, 18 years ago

ajout mesures pour PbG4 pour compilation avec -O2 (au lieu de -g), Reza 08/05/2007

File size: 17.8 KB
Line 
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
9 Flags de compilation avec [-O -g]
10(c) sgsda: AMD Bipro AMD opteron 248 @ 2.2 GHz (amd-lx-
11 Flags de compilation avec [-O -g]
12(cc) grid-saclay: AMD opteron 275 Bipro-bicoeur @ 2.2 GHz (amd275-lx)
13 Flags de compilation avec [-O -g]
14
15(d) asc: bipro alpha (@ ~1 GHz) server DS20 OSF (osf1) , cxx 6.5 (osf-asc)
16(nouveau asc 420 MFLOPS, moins puissante que l'ancien asc 800 MFLOPS)
17(e) xp1000-dapnia: alpha xp1000 @ ~ 600 MHz ? OSF1 , cxx ? (osf-xp1000)
18(e') cool: alpha xp1000 @ ~ 667 MHz ? OSF1 5.1 , cxx 6.3
19(f) superosf-dapnia: multi-proc alphaServer ES80 6 procs EV7 @ 1 GHz (super-osf)
20
21(g) ccsvx01: XServe G5 bipro @~1GHz (Darwin/OSX) (G5-osx-1GHz) , gcc 3.3
22(h) PowerBook-Reza : Apple G4 @ 1.25 GHz (G4-osx-1.25GHz) , gcc 3.3
23(i) MacBook-Reza: Apple/ Core double-coeur Intel @ 1.83 GHz (core-osx-1.83GHz) gcc 4
24(j) MacPro-Grosdidier : Apple / Xeon 2 double-coeur @ 3 GHz gcc 4.0.1 , compil SOPHYA -O2 -g
25
26(p) IBM-AIX regatta , xlC , IBM eServer pSeries 655 , 8 proc power4 @ 1.1 GHz
27(q) IBM-AIX meso , AIX 5.3, xlC V8 , IBM Power5 , 8 proc bi-coeur P575 @ 1.9 GHz
28
29(s) SGI-IRIX64 magique, CC
30
31NOTES :
32- Sur les machines Xeon, il y a une interaction entre process / threads par rapport a
33l'occupation des CPU's. On perd un facteur 3 en performance multi-threads/multi-taches.
34La machine MacPro avec OSX se debrouille quand meme mieux.
35- Effet du systeme ou carte mere ???
36
37Flag de compilation
38- Flag de compilation par defaut [-O -g] en general
39- Sur eros3 (xeon-linux gcc 3.3) [-O -g] OU [-O3 -g]
40- Sur Darwin [-g] ou [-O2 -g] (ou [-tune G5] sur XServe G5)
41 Sur les mac (en particulier G4/G5), grande difference entre -g et -Ox -g
42 mais peu de difference entre -O -O2 -O3
43- Sur machine aix-meso [-O -g] ou [-O3 -g]
44
45X/ Donnees SPECint2000 (3) / SPECfp2000 (2) (http//www.spec.org)
46
47(1) MFLOPS -> cpupower 2 (x/y : -O -g / -O3)
48----------------------------------------------------------------------
49 MFLOPS(1) SPECfp SPECint
50----------------------------------------------------------------------
51(b)xeon-lx-2.8GHz 900 1400 1400
52(c)amd-lx 690 1600 1300
53(cc)amd2-lx 675 1600 1300
54
55(d)osf-xp1000 648 500 400
56(f)superosf 842 1100 700
57
58(f)G4-osx-1.25GHz 92/380 (-g)/(-O2 -g)
59(i)core-osx-1.83GHz 310/880 1400 1500 (-g)/(-O2 -g)
60(j)xeon-osx 2600 2900 -
61
62(p)ibm-aix-regatta 730/1750 1050 700 (-O -g)/(-O3)
63(p)ibm-aix-meso 1250/3600 (-O -g)/(-O3)
64----------------------------------------------------------------------
65
66
67A/ Compilation tout SOPHYA
68----------------------------
69csh> time make all (1)
70ou
71csh> time make -j 2 all (2)
72 Temps CPU
73 Indice de performance 100*(1000/TCPU)
74 Temps elapsed (vrai)
75 Temps vrai / TCPU
76
77
78----------------------------------------------------------------------
79 CPU(s) IndPerf TElapsed , TCPU/Elapsed %
80----------------------------------------------------------------------
81(a)xeon-lx-2.4GHz (2) 615 s 162 410 s 150%
82 avec -O3 -g (2) 1300 s 77 760 s 172%
83(b)xeon-lx-2.8GHz (2) 755 s 132 540 s 140%
84(c)amd-lx (2) 336 s 297 175 s 192%
85
86(d)osf-asc (1) 1920 s 52 2340 s 83% (??)
87(e)osf-xp1000 (1) 533 s 187 660 s 80%
88(f)superosf (1) 895 s 112 910 s 98%
89
90(f)G4-osx-1.25GHz (1) 660 s 151 710 s 93% [-g]
91 1500 s 94% [-O2 -g]
92(h)G5-osx-1GHz (2) 453 s 221 250 s 182%
93 -tune=G5 1100 s 90
94(i)core-osx-1.83GHz (2) 209 s 478 116 s 180%
95 -O2 (1) 367 s 272 381 96%
96(j)xeon-osx
97
98(p)ibm-aix
99----------------------------------------------------------------------
100
101Taille shared libs :
102(a)
103(f) = (e) = 57 MO
104(h) 80 MO
105(i) 83 MO
106
107B/ Calcul brut (Tableaux de SOPHYA) avec / sans threads
108--------------------------------------------------------
109
110B.1/ arr = c1*a1+c2*a2 ( ~ 3 10^6 op. double)
111(1) time cpupower 2 # compile avec -O3 (/ -O -g)
112(2) time zthr arr 1 1000 1 thread
113(3) time zthr arr 2 1000 2 thread
114(4) time zthr arr 4 1000 4 thread
115(5) time zthr arr 6 1000 6 thread
116(6) time zthr arr 8 1000 8 thread
117
118-----------------------------------------------------------------------------------
119 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
120-----------------------------------------------------------------------------------
121(a)xeon-lx-2.4GHz 1167 5.15/5.2/99% 11.4/5.8/196%
122 (4)36.6/9.28/394%
123 -O3 -g 4.9/5./99%
124(b)xeon-lx-2.8GHz 920 2.3/2.3/100% 6.2/3.1/198%
125 (4)26/6.6/396%
126(c)amd-lx 690 3.6/3.6/99% 6.8/4/171%
127 (4)13.5/7/193%
128 (5)20.3/10.23/198%
129(cc)amd2-lx 675 2/2/99% 4.15/2.1/197%
130 (4)8.25/4.15/198%
131 (5)13.6/4.6/292%
132 (6)19.8/6.5/300%
133
134(d)osf-asc 420 6.3s/6.5s/99% 16.9/8.8/192%
135 (4)29.9/15.7/191%
136(e)osf-xp1000 648 5.1/5.3/96.6% 11.4/11.4/99%
137 (4)25.2/25.5/99%
138(f)superosf 842 2.87/2.88/99.6% 6.25/4.1/153%
139 (4)11.6/3.06/379%
140
141(f)G4-osx-1.25GHz 92 44s/48s/91% 86.7/99.8/92% [-g]
142 380 12.2/12.9/95% 24/25.3/95% [-O2 -g]
143(h)G5-osx-1GHz 1151 20s/20s/99% 40s/23s/170%
144 (4) 80.8/45/180%
145 -tune=G5 3.35/3.8/88% 7.1/3.6/196%
146 (4) 14/7.5/187%
147(i)core-osx-1.83GHz 855 11.5/11.5/100% 23/11.6/192% [-g]
148 (4) 46/23/199%
149 -O2 3.85/3.89/99% 7.7/3.9/198% [-O2 -g]
150 (4) 15.4/7.77/198%
151
152(j)xeon-osx 2600 2.5/2.5/100% 5.1/2.6/199%
153 (4) 11.5/3.2/362%
154 (5) 17.4/4.77/365%
155
156(p)ibm-aix-regatta 1750/730 6.8/6.9/98% 13.1/6.75/195%
157 (4) 26.3/11.7/225%
158(q)ibm-aix-meso 3600/1250 3.6/3.75/96% 7.35/3.7/197%
159 (4) 12.46/4.2/298%
160 (5) 219/6.7/280%
161 (6) 24/4.5/530%
162
163
164(s)sgi-magique 460 60/60/99%
165 -----------------------------------------------------------------------------------
166
167
168B.2/ Multiplication de matrices mtx = mtx1 * mtx2
169 ~ 2 10^9 op. double / thread
170(1) time cpupower 2 (-O3 / -O -g)
171(2) time zthr mtx 1 1000 1 thread
172(3) time zthr mtx 2 1000 2 thread
173(4) time zthr mtx 4 1000 4 thread
174(5) time zthr mtx 6 1000 6 thread
175(6) time zthr mtx 8 1000 8 thread
176
177-----------------------------------------------------------------------------------
178 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
179-----------------------------------------------------------------------------------
180(a)xeon-lx-2.4GHz 1167 6.5/6.5/100% 17.4/8.8/198%
181 (4) 80.5/20.3/397%
182 (5) 114.5/29.6/387%
183 (6) 160/40.3/387%
184
185(b)xeon-lx-2.8GHz 920 3.4/3.4/100% 12/6.1/199%
186 (4) 55.8/14/400%
187(c)amd-lx 690 6.98/6.98/100% 14.1/8.15/173%
188 (4) 27.7/14.23/194%
189 (5) 41.4/21.07/196%
190 (6) 55.4/27.9/198.7%
191(cc)amd2-lx 675 4.1/4.1/100% 9.55/4.8/198%
192 (4) 20/10.27/195%
193 (5) 32.8/11.16/294%
194 (6) 42.75/13.8/309%
195
196
197(d)osf-asc 420 13.5s/13.7s/98% 32/16.5/194%
198 (4) 67.5/34.4/196%
199(e)osf-xp1000 648 13/14.1/92% 27.1/27.4/99%
200 (4) 54/54.7/99.6%
201 (5) 80.6/81/99.6%
202 (6) 107.8/108.3/99.5%
203(f)superosf 842 6.1/7.24/84% 12.35/6.29/196%
204 (4) 24.3/6.31/385%
205 (5) 36.5/10.9/335%
206 (6) 50.1/18.15/276%
207
208(f)G4-osx-1.25GHz 333 23.5/24.5/96% [-O2]
209(h)G5-osx-1GHz 1151 23/23.7/97% 46.5/27.5/170%
210 (4) 93.4/49.4/189%
211 -tune=G5 5.7/5.8/98% 13.3/6.8/197%
212 (4) 26.8/13.56/197%
213 (6) 53.8/27.25/197%
214(i)core-osx-1.83GHz 855 12.6/12.7/100% 25.8/13.4/194%
215 (4) 51.6/26/199%
216 -O2 4.25/4.5/94% 10.6/5.36/198%
217 (4) 20.87/10.68/198%
218 -O2 2 jobs // 2 x 5/5.4/92%
219(j)xeon-osx 2600 2.8/2.8/99% 9.3/4.66/199%
220 (4) 31.4/8.6/364%
221 (5) 47.1/12.96/364%
222 (6) 62.8/17.38/362%
223
224(p)ibm-aix-regatta 1750/730 9.5/9.7/98% 18.3/16.0/114%
225 (4) 38.3/24.7/155%
226(p)ibm-aix-meso 3600/1250 2.3/2.3/99% 5.1/2.64/194% (compil avec -O3)
227 (4) 11.4/4.16/272%
228 (5) 20.2/5.85/344%
229 (6) 29.9/6.74/442%
230
231(s)sgi-magique 460 49/49/99% 101/56/181%
232
233 -----------------------------------------------------------------------------------
234
235C/ Calcul fft (FFTW , FFTPack )
236-------------------------------
237
238(1) time cpupower 2
239(2) time tfft 2000000 W d 0 0 (avec FFTW)
240(3) time tfft 2000000 P d 0 0 (avec FFTPack_Sophya)
241
242IndPerf=1000/TCPU
243
244
245-----------------------------------------------------------------------------------
246 (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf
247-----------------------------------------------------------------------------------
248(a)xeon-lx-2.4GHz 1167 5.5/5.6/97% 180 7.4/7.4/100% 135
249 -O3 -g 6.8/7.1/96% 147
250(b)xeon-lx-2.8GHz 920
251(c)amd-lx 690 3.2/3.75/86% 4.2/4.2/100% 238
252 ~2x 4.7/4.7/99%
253(cc)amd2-lx 675 2.8/2.8/99% 3.56/3.58/99%
254
255(d)osf-asc 420 13.3/13.9/95.7% 75 12.2/17.6/70% 82
256(e)osf-xp1000 648 9.9/10.2/97% 101 9.3/9.46/98.5% 107
257(e')cool 9.1/9.3/98%
258(f)superosf 842 6./6.22/96.7% 166 5.1/5.18/97.4% 196
259
260(f)G4-osx-1.25GHz 92 15.2/15.9/96% 66 23.8/34.1/70% 42 [-g]
261 380 8.8/10.3/86% 14.7/20.9/65% [-O2 -g]
262(h)G5-osx-1GHz 1151 8.5/8.7/97.5% 120 11.5/11.6/99% 87
263 -tune=G5 5/5.1/98% 200 5.5/5.6/97% 180
264(i)core-osx-1.83GHz 855 4.6/4.75/97% 217 7/7.08/99% 142
265 -O2 3.2/3.2/99% 312 3.6/3.6/99% 277
266 -O2 2 jobs // 2 x 3.9/4.3/90% 2x 4.7/5/91% 200
267(j)xeon-osx 2600 2.6/2.6/98% 384
268 2 jobs // ~2x 3.6/4.1/87% 250
269 4 jobs //
270
271(p)ibm-aix-regatta 1750/730 6.25/18.9/33% 160 5.25/15.7/33% 190
272(q)ibm-aix-meso 3600/1250 3.95/4.3/91% 250 3.82/4./94% 260
273 2 jobs // ~2x 3.88/4.2/92% 250
274
275
276(s)sgi-magique 460 22/22/98% 42 24.5/25/99% 40
277 -----------------------------------------------------------------------------------
278
279
280D/ Calcul inversion par lapack
281-------------------------------
282
283lpk inverse 1000,1000 0
284---> temps de calcul inversion par lapack
285-------------------------------------------------------------------------------------------
286 CPU/Elap/% (1)
287-------------------------------------------------------------------------------------------
288(a)xeon-lx-2.4GHz 5.6/~100%
289(b)xeon-lx-2.8GHz
290(c)amd-lx 5.5/5.5/99%
291
292(d)osf-asc
293(e')cool 2.8/2.9/95%
294(f)superosf
295
296(f)G4-osx-1.25GHz 2.3/~100% [-O2 -g]
297(h)G5-osx-1GHz 0.8/~100%
298 -tune=G5
299(i)core-osx-1.83GHz 2/~100%
300 -O2
301(p)ibm-aix-regatta
302(q)ibm-aix-meso 0.55/~100%
303
304--------------------------------------------------------------------------------------------
305
306
307K/ Efficacite de gestion de lock (mutex) avec les threads et tableaux
308----------------------------------------------------------------------
309(32 threads - operant sur 2000 vecteurs ~ 64000 lock/unlock/wait/broadcast)
310
311(1) time zthr syncp 32 2000 4
312(2) time zthr sync 32 2000 4
313(1) time zthr syncp 4 15000 130
314(2) time zthr sync 4 15000 130
315-------------------------------------------------------------------------------------------
316 CPU/Elap/% (1) (2) (3) (4)
317-------------------------------------------------------------------------------------------
318(a)xeon-lx-2.4GHz 23.5/14/168% 4.3/1.2/365% 7.9/5.5/142% 4/2.15/190%
319 Avant ThSafeOp 17/178%
320 avec -O3 -g
321(b)xeon-lx-2.8GHz (2)
322(c)amd-lx 0.6/1/63% 0.6/1/60% 3.5/3.5/102% 2.6/2.7/98%
323
324(d)osf-asc 4.5/3.4/132% 3.35/2/170% 15.8/10.5/150% 13/8/163%
325 5.4/100%(NoThSafe)
326(e')cool 1.3/1.37/95% 1.35/1.5/89% 5.3/5.3/99% 5.2/5.2/99%
327(e)superosf (1)
328(
329(f)G4-osx-1.25GHz (1) 40.5/42.6/95% 42.2/43.5/97% [-g]
330 3.9/4.3/89% 3.5/4/89% 3.8/4/95% 4.3/4.6/93%
331
332(h)G5-osx-1GHz (2) 2.6/130% (NoThSafe)
333 -tune=G5
334(i)core-osx-1.83GHz 7.7/7.1/108% 7.8/6.7/116% 30.2/29.6/102% 30.3/29.2/104%
335 -O2
336(j)xeon-osx
337 Avant ThSafeOp 2.55/143%
338
339(p)ibm-aix-regatta 4.7/111%
340(q)ibm-aix-meso 7.5/2.8/300% 17/3.8/450% 8.2/3.05/270% 4.85/2.43/200%
341
342--------------------------------------------------------------------------------------------
343
344
345
346L/ I/O et PPF
347-----------------
348Ecriture/lecture de n=10^7 lignes de int+6double, Total ~ 500 MO
349(1) time tstdtable w xx.ppf swap 10000000 1024 0
350(2) time tstdtable r xx.ppf swap 10000000 1024 0
351(3) time tstdtable w xx.ppf swap 50000000 1024 0
352(4) time tstdtable r xx.ppf swap 50000000 1024 0
353
354-------------------------------------------------------------------------------------------
355 CPU/Elap/% (1) (2) (3) (4)
356-------------------------------------------------------------------------------------------
357(a)xeon-lx-2.4GHz 17/26/63% 5.5/5.6/94%
358(b)xeon-lx-2.8GHz
359(c)amd-lx 5.9/6./97% 3.4/3.4/100% 30/32/93% 24/165/13% ?
360
361(d)osf-asc
362(e')cool 15/30/50% 13/13/99%
363(f)superosf
364
365(f)G4-osx-1.25GHz 26/29.7/87% 15.7/38.6/41% [-O2 -g]
366(h)G5-osx-1GHz
367 -tune=G5
368(i)core-osx-1.83GHz 37/37.8/98% 22/48.4/45%
369 -O2
370(p)ibm-aix-regatta
371(q)ibm-aix-meso 5.5/16.8/38% 5.7/13.2/43% 32.7/85/39% 29/60/49%
372
373--------------------------------------------------------------------------------------------
Note: See TracBrowser for help on using the repository browser.