Benchmark Results of Tempest Beowulf Cluster
Benchmark Results
NAS Benchmark:
|
Benchmark Name
|
Number of nodes
|
Mpich Results(Mflops/sec)
|
LAM Results(Mflops/sec)
|
LAM with Fujitsu compiler
|
T3E 900 at PSC
|
|
BT class=A
|
9
|
382.36(440.12 sec)
|
438.53(383.75 sec)
|
578.13(291.08 sec)
|
534.96(314.57 sec)
|
|
IS class=A
|
8
|
4.64(18.06 sec)
|
13.17(6.37 sec)
|
14.35(5.85 sec)
|
24.78(3.39 sec)
|
|
CG class=A
|
8
|
123.69(12.10 sec)
|
136.96(10.93 sec)
|
178.12(8.40 sec)
|
197.11(7.59 sec)
|
|
EP class=A
|
9
|
15.65(34.31 sec)
|
17.37(30.91 sec)
|
24.33(22.07 sec)
|
NA
|
|
LU class=A
|
8
|
432.07(276.10 sec)
|
417.79(285.54 sec)
|
573.92(207.86 sec)
|
514.82(231.72 sec)
|
|
MG class=A
|
8
|
298.09(13.06 sec)
|
338.70(11.49 sec)
|
439.42(8.86 sec)
|
584.54(6.66 sec)
|
|
SP class=A
|
9
|
262.69(323.61 sec)
|
266.66(318.80 sec)
|
329.34(258.12 sec)
|
354.01(240.14 sec)
|
|
FT class=A
|
8
|
NA
|
NA
|
260.43(27.40 sec)
|
NA
|
compiler options: g77 -O3 -mpentiumpro -ffast-math -funroll-loops -malign-double
Fujitsu compiler options: frt -Kfast,eval,PENTIUM_PRO,fastlib -x-
T3E cf90 options: cf90 -O 3
|
Benchmark Name
|
Number of Nodes
|
Tempest With Fujitsu Compiler
|
CRAY T3E900(from NAS,NASA)
|
|
BT class=A
|
16
|
853.82 MFlops(197.1 sec)
|
879.2 Mflops(191.4 sec)
|
|
IS class=A
|
16
|
15.71 Mflops(5.34 sec)
|
35 Mflops(2.4 sec)
|
|
CG class=A
|
16
|
261.56 Mflops(6.91 sec)
|
299.3 Mflops(5 sec)
|
|
EP class=A
|
16
|
43.04 Mflops(12.47 sec)
|
41.6 Mflops(12.9 sec)
|
|
LU class=A
|
16
|
1154.9 Mflops(103.3 sec)
|
1022.2 Mflops(116.7 sec)
|
|
MG class=A
|
16
|
758.74 Mflops(5.13 sec)
|
1255.6 Mflops(3.1 sec)
|
|
SP class=A
|
16
|
546.12 Mflops(155.66 sec)
|
643 Mflops(132.2 sec)
|
|
FT class=A
|
16
|
566.63 Mflops(12.59 sec)
|
648.8Mflops(11 sec)
|
|
Benchmark Name
|
Number of Nodes
|
Tempest With Fujitsu Compiler
|
CRAY T3E900(from NAS,NASA)
|
|
BT class=B
|
16
|
1097.19 Mflops(639.98 sec)
|
902.5 Mflops(778 sec)
|
|
IS class=B
|
16
|
25.84 Mflops(12.99 sec)
|
32.3 Mflops(10.4 sec)
|
|
CG class=B
|
16
|
198.74 Mflops(275.28 sec)
|
164.6 Mflops(332.4 sec)
|
|
EP class=B
|
16
|
43.06 Mflops(49.87 sec)
|
41.7 Mflops(51.5 sec)
|
|
LU class=B
|
16
|
1175 .16 Mflops(424.47 sec)
|
1005.1 Mflops(496.3sec)
|
|
MG class=B
|
16
|
824.08 Mflops(23.6 sec)
|
1323.9 Mflops(14.7 sec)
|
|
SP class=B
|
16
|
630.59 Mflops(562.98 sec)
|
681.4 Mflops(521 sec)
|
|
FT class=B
|
16
|
550.12 Mflops(167.33 sec)
|
545.3 Mflops(168.8 sec)
|
|
Benchmark Name
|
Number of Nodes
|
Tempest With Fujitsu Compiler
|
CRAY T3E1200(from NAS,NASA)
|
|
BT class=C
|
16
|
1250.88 Mflops(2291.41 sec)
|
1088.3 Mflops(2633.8 sec)
|
|
IS class=C
|
16
|
28.83 Mflops(46.55 sec)
|
26.9 Mflops(49.8 sec)
|
|
CG class=C
|
16
|
211.63 Mflops(677.34 sec)
|
148.7 Mflops(963.8 sec)
|
|
EP class=C
|
16
|
52.56 Mflops(163.44 sec)
|
54.8 Mflops(156.9 sec)
|
|
LU class=C
|
16
|
1380 .13 Mflops(1477.4 sec)
|
1169.5 Mflops(1743.5 sec)
|
|
MG class=C
|
16
|
1071.59 Mflops(145.29 sec)
|
1616.8 Mflops(96.3 sec)
|
|
SP class=C
|
16
|
871.19 Mflops(1664.52 sec)
|
800.4 Mflops(1811.7 sec)
|
|
FT class=C
|
16
|
392.64 Mflops(1009.55 sec)
|
NA
|
High Performance LINPACK (HPL + ATLAS, using 8, 10 and 16 CPUs):
|
Matrix Size
|
Block size
|
P
|
Q
|
Time
|
Results(Gflops)
|
|
20000
|
200
|
2
|
4
|
1501 sec
|
3.55
|
|
25000
|
200
|
2
|
5
|
2324 sec
|
4.48
|
|
34000
|
200
|
4
|
4
|
3545 sec
|
7.4
|
compiler options: g77 -O3 -mpentiumpro -ffast-math -funroll-loops -malign-double
More benchmark results will be available soon!