You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Simon Chow <si...@gmail.com> on 2008/03/07 13:39:36 UTC
[general][evaluation] I did a performance evaluation using Scimark2
I use a scientific computing benchmark Scimark2, which has 2 running modes:
default and -large.
I would like to share it with you. :=)
Platform:
Intel(R) Xeon(TM) CPU 2.80GHz*4.
arch: x86
os: Linux 2.6.18-8.el5xen;
Mem:4GB
Harmony
-Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline
Composite Score
FFT
(1024)
SOR
(100*100)
Monte Carlo
Sparse matmult
(N=1000,nz=5000)
LU
(100*100)
193.99
223.91
366.62
28.42
184.19
166.83
194.05
222.20
370.43
28.04
183.16
166.42
193.67
223.05
369.72
28.61
181.29
165.70
193.41
221.29
371.28
27.69
182.04
164.74
194.34
222.48
371.00
28.17
183.32
166.75
-Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline -large
Composite Score
FFT
(1048576)
SOR
(1000*1000)
Monte Carlo
Sparse matmult
(N=100000,
nz=1000000)
LU
(1000*1000)
179.31
37.93
359.34
27.18
289.51
182.60
178.31
35.84
359.34
28.08
288.78
179.50
179.35
37.19
258.66
28.08
289.43
183.40
179.02
35.63
360.01
27.14
289.92
182.40
179.80
37.44
360.01
27.25
290.08
184.21
Sun sdk1.5
-Xms1500m -Xmx1500m -server jnt.scimark2.commandline
Composite Score
FFT
(1024)
SOR
(100*100)
Monte Carlo
Sparse matmult
(N=1000,nz=5000)
LU
(100*100)
427.30
252.57
593.82
22.51
321.41
946.18
431.48
272.11
596.21
22.16
322.68
944.21
432.80
273.99
596.77
22.54
322.20
948.48
428.75
256.96
596.03
22.58
323.63
944.54
432.90
276.25
597.32
22.59
323.16
945.19
-Xms1500m -Xmx1500m –server jnt.scimark2.commandline -large
Composite Score
FFT
(1048576)
SOR
(1000*1000)
Monte Carlo
Sparse matmult
(N=100000,
nz=1000000)
LU
(1000*1000)
243.25
36.42
553.20
34.72
381.71
265.18
278.28
37.74
576.72
39.89
369.94
367.11
266.89
37.42
575.21
41.22
368.48
312.11
271.74
37.63
577.16
39.48
371.28
333.17
269.53
37.49
574.99
41.12
368.88
325.20
gcj-4.0.2 –O3
Composite Score
FFT
(1024)
SOR
(100*100)
Monte Carlo
Sparse matmult
(N=1000,
nz=5000)
LU
(100*100)
214.69
228.30
360.18
11.19
151.84
321.94
220.42
195.46
338.18
7.96
276.17
284.33
254.33
214.59
360.18
11.58
277.23
408.05
179.55
184.54
355.71
6.71
143.22
227.56
233.90
215.02
360.58
11.57
276.41
305.92
-large
Composite Score
FFT
(1048576)
SOR
(1000*1000)
Monte Carlo
Sparse matmult
(N=100000,
nz=1000000)
LU
(1000*1000)
192.24
29.62
348.23
11.55
222.95
348.86
177.07
35.24
322.72
8.16
232.94
286.25
174.29
35.02
331.95
9.75
249.63
245.09
196.79
27.28
347.29
11.50
255.12
342.76
179.69
37.69
349.346
10.69
176.19
324.57
--
>From : Simon.Chow@Software School of Fudan University
Re: [general][evaluation] I did a performance evaluation using Scimark2
Posted by Aleksey Shipilev <al...@gmail.com>.
Simon, all,
You also might look into the issue [1] I've found recently, it speeds
up MonteCarlo 2 times.
Thanks,
Aleksey.
[1] https://issues.apache.org/jira/browse/HARMONY-5584
On Fri, Mar 7, 2008 at 6:25 PM, Aleksey Shipilev
<al...@gmail.com> wrote:
> Wow, Simon :)
>
> Can you align this data? It's completely unreadable - I haven't clue
> how Harmony performs looking to these numbers. I'm very interested in
> this measurements.
>
> Thanks,
> Aleksey.
>
Re: [general][evaluation] I did a performance evaluation using Scimark2
Posted by Simon Chow <si...@gmail.com>.
I am sorry about that...
In my Gmail browser, they are some pretty good-looking tables
maybe Gmail don't support table well :(
So this is the plain text version ( and a snapshot.jpeg with some content is
in attachment )
Platform:
Intel(R) Xeon(TM) CPU 2.80GHz*4.
Linux localhost 2.6.18-8.el5xen;
Mem:4GB
Harmony (M5)
-Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline
Composite Score FFT(1024) SOR(100*100) Monte Carlo Sparse
matmult(N=1000,nz=5000) LU(100*100)
193.99 223.91 366.62 28.42 184.19 166.83
194.05 222.20 370.43 28.04 183.16 166.42
193.67 223.05 369.72 28.61 181.29 165.70
193.41 221.29 371.28 27.69 182.04 164.74
194.34 222.48 371.00 28.17 183.32 166.75
-Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline -large
Composite Score FFT(1048576) SOR(1000*1000) Monte Carlo Sparse
matmult(N=100000,nz=1000000) LU(1000*1000)
179.31 37.93 359.34 27.18 289.51 182.60
178.31 35.84 359.34 28.08 288.78 179.50
179.35 37.19 258.66 28.08 289.43 183.40
179.02 35.63 360.01 27.14 289.92 182.40
179.80 37.44 360.01 27.25 290.08 184.21
Sun java version "1.5.0_12"
-Xms1500m -Xmx1500m -server jnt.scimark2.commandline
Composite Score FFT(1024) SOR(100*100) Monte Carlo Sparse
matmult(N=1000,nz=5000) LU(100*100)
427.30 252.57 593.82 22.51 321.41 946.18
431.48 272.11 596.21 22.16 322.68 944.21
432.80 273.99 596.77 22.54 322.20 948.48
428.75 256.96 596.03 22.58 323.63 944.54
432.90 276.25 597.32 22.59 323.16 945.19
-Xms1500m -Xmx1500m –server jnt.scimark2.commandline -large
Composite Score FFT(1048576) SOR(1000*1000) Monte Carlo Sparse
matmult(N=100000,nz=1000000) LU(1000*1000)
243.25 36.42 553.20 34.72 381.71 265.18
278.28 37.74 576.72 39.89 369.94 367.11
266.89 37.42 575.21 41.22 368.48 312.11
271.74 37.63 577.16 39.48 371.28 333.17
269.53 37.49 574.99 41.12 368.88 325.20
gcj-4.0.2 –O3
Composite Score FFT(1024) SOR(100*100) Monte Carlo Sparse
matmult(N=1000,nz=5000) LU(100*100)
214.69 228.30 360.18 11.19 151.84 321.94
220.42 195.46 338.18 7.96 276.17 284.33
254.33 214.59 360.18 11.58 277.23 408.05
179.55 184.54 355.71 6.71 143.22 227.56
233.90 215.02 360.58 11.57 276.41 305.92
-large
Composite Score FFT(1048576) SOR(1000*1000) Monte Carlo Sparse
matmult(N=100000,nz=1000000) LU(1000*1000)
192.24 29.62 348.23 11.55 222.95 348.86
177.07 35.24 322.72 8.16 232.94 286.25
174.29 35.02 331.95 9.75 249.63 245.09
196.79 27.28 347.29 11.50 255.12 342.76
179.69 37.69 349.346 10.69 176.19 324.57
On 07/03/2008, Aleksey Shipilev <al...@gmail.com> wrote:
>
> Wow, Simon :)
>
> Can you align this data? It's completely unreadable - I haven't clue
> how Harmony performs looking to these numbers. I'm very interested in
> this measurements.
>
> Thanks,
>
> Aleksey.
>
>
> On Fri, Mar 7, 2008 at 3:39 PM, Simon Chow <si...@gmail.com>
> wrote:
> > I use a scientific computing benchmark Scimark2, which has 2 running
> modes:
> > default and -large.
> > I would like to share it with you. :=)
> >
> >
> > Platform:
> > Intel(R) Xeon(TM) CPU 2.80GHz*4.
> > arch: x86
> > os: Linux 2.6.18-8.el5xen;
> > Mem:4GB
> >
> > Harmony
> >
> > -Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline
> >
> > Composite Score
> >
> > FFT
> >
> > (1024)
> >
> > SOR
> >
> > (100*100)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=1000,nz=5000)
> >
> > LU
> >
> > (100*100)
> >
> > 193.99
> >
> > 223.91
> >
> > 366.62
> >
> > 28.42
> >
> > 184.19
> >
> > 166.83
> >
> > 194.05
> >
> > 222.20
> >
> > 370.43
> >
> > 28.04
> >
> > 183.16
> >
> > 166.42
> >
> > 193.67
> >
> > 223.05
> >
> > 369.72
> >
> > 28.61
> >
> > 181.29
> >
> > 165.70
> >
> > 193.41
> >
> > 221.29
> >
> > 371.28
> >
> > 27.69
> >
> > 182.04
> >
> > 164.74
> >
> > 194.34
> >
> > 222.48
> >
> > 371.00
> >
> > 28.17
> >
> > 183.32
> >
> > 166.75
> >
> > -Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline -large
> >
> > Composite Score
> >
> > FFT
> >
> > (1048576)
> >
> > SOR
> >
> > (1000*1000)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=100000,
> >
> > nz=1000000)
> >
> > LU
> >
> > (1000*1000)
> >
> > 179.31
> >
> > 37.93
> >
> > 359.34
> >
> > 27.18
> >
> > 289.51
> >
> > 182.60
> >
> > 178.31
> >
> > 35.84
> >
> > 359.34
> >
> > 28.08
> >
> > 288.78
> >
> > 179.50
> >
> > 179.35
> >
> > 37.19
> >
> > 258.66
> >
> > 28.08
> >
> > 289.43
> >
> > 183.40
> >
> > 179.02
> >
> > 35.63
> >
> > 360.01
> >
> > 27.14
> >
> > 289.92
> >
> > 182.40
> >
> > 179.80
> >
> > 37.44
> >
> > 360.01
> >
> > 27.25
> >
> > 290.08
> >
> > 184.21
> >
> >
> > Sun sdk1.5
> >
> > -Xms1500m -Xmx1500m -server jnt.scimark2.commandline
> >
> > Composite Score
> >
> > FFT
> >
> > (1024)
> >
> > SOR
> >
> > (100*100)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=1000,nz=5000)
> >
> > LU
> >
> > (100*100)
> >
> > 427.30
> >
> > 252.57
> >
> > 593.82
> >
> > 22.51
> >
> > 321.41
> >
> > 946.18
> >
> > 431.48
> >
> > 272.11
> >
> > 596.21
> >
> > 22.16
> >
> > 322.68
> >
> > 944.21
> >
> > 432.80
> >
> > 273.99
> >
> > 596.77
> >
> > 22.54
> >
> > 322.20
> >
> > 948.48
> >
> > 428.75
> >
> > 256.96
> >
> > 596.03
> >
> > 22.58
> >
> > 323.63
> >
> > 944.54
> >
> > 432.90
> >
> > 276.25
> >
> > 597.32
> >
> > 22.59
> >
> > 323.16
> >
> > 945.19
> >
> >
> > -Xms1500m -Xmx1500m –server jnt.scimark2.commandline -large
> >
> > Composite Score
> >
> > FFT
> >
> > (1048576)
> >
> > SOR
> >
> > (1000*1000)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=100000,
> >
> > nz=1000000)
> >
> > LU
> >
> > (1000*1000)
> >
> > 243.25
> >
> > 36.42
> >
> > 553.20
> >
> > 34.72
> >
> > 381.71
> >
> > 265.18
> >
> > 278.28
> >
> > 37.74
> >
> > 576.72
> >
> > 39.89
> >
> > 369.94
> >
> > 367.11
> >
> > 266.89
> >
> > 37.42
> >
> > 575.21
> >
> > 41.22
> >
> > 368.48
> >
> > 312.11
> >
> > 271.74
> >
> > 37.63
> >
> > 577.16
> >
> > 39.48
> >
> > 371.28
> >
> > 333.17
> >
> > 269.53
> >
> > 37.49
> >
> > 574.99
> >
> > 41.12
> >
> > 368.88
> >
> > 325.20
> >
> >
> > gcj-4.0.2 –O3
> >
> > Composite Score
> >
> > FFT
> >
> > (1024)
> >
> > SOR
> >
> > (100*100)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=1000,
> >
> > nz=5000)
> >
> > LU
> >
> > (100*100)
> >
> > 214.69
> >
> > 228.30
> >
> > 360.18
> >
> > 11.19
> >
> > 151.84
> >
> > 321.94
> >
> > 220.42
> >
> > 195.46
> >
> > 338.18
> >
> > 7.96
> >
> > 276.17
> >
> > 284.33
> >
> > 254.33
> >
> > 214.59
> >
> > 360.18
> >
> > 11.58
> >
> > 277.23
> >
> > 408.05
> >
> > 179.55
> >
> > 184.54
> >
> > 355.71
> >
> > 6.71
> >
> > 143.22
> >
> > 227.56
> >
> > 233.90
> >
> > 215.02
> >
> > 360.58
> >
> > 11.57
> >
> > 276.41
> >
> > 305.92
> >
> > -large
> >
> > Composite Score
> >
> > FFT
> >
> > (1048576)
> >
> > SOR
> >
> > (1000*1000)
> >
> > Monte Carlo
> >
> > Sparse matmult
> >
> > (N=100000,
> >
> > nz=1000000)
> >
> > LU
> >
> > (1000*1000)
> >
> > 192.24
> >
> > 29.62
> >
> > 348.23
> >
> > 11.55
> >
> > 222.95
> >
> > 348.86
> >
> > 177.07
> >
> > 35.24
> >
> > 322.72
> >
> > 8.16
> >
> > 232.94
> >
> > 286.25
> >
> > 174.29
> >
> > 35.02
> >
> > 331.95
> >
> > 9.75
> >
> > 249.63
> >
> > 245.09
> >
> > 196.79
> >
> > 27.28
> >
> > 347.29
> >
> > 11.50
> >
> > 255.12
> >
> > 342.76
> >
> > 179.69
> >
> > 37.69
> >
> > 349.346
> >
> > 10.69
> >
> > 176.19
> >
> > 324.57
> >
> >
> >
> > --
> > From : Simon.Chow@Software School of Fudan University
> >
>
--
>From : Simon.Chow@Software School of Fudan University
Re: [general][evaluation] I did a performance evaluation using Scimark2
Posted by Aleksey Shipilev <al...@gmail.com>.
Wow, Simon :)
Can you align this data? It's completely unreadable - I haven't clue
how Harmony performs looking to these numbers. I'm very interested in
this measurements.
Thanks,
Aleksey.
On Fri, Mar 7, 2008 at 3:39 PM, Simon Chow <si...@gmail.com> wrote:
> I use a scientific computing benchmark Scimark2, which has 2 running modes:
> default and -large.
> I would like to share it with you. :=)
>
>
> Platform:
> Intel(R) Xeon(TM) CPU 2.80GHz*4.
> arch: x86
> os: Linux 2.6.18-8.el5xen;
> Mem:4GB
>
> Harmony
>
> -Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline
>
> Composite Score
>
> FFT
>
> (1024)
>
> SOR
>
> (100*100)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=1000,nz=5000)
>
> LU
>
> (100*100)
>
> 193.99
>
> 223.91
>
> 366.62
>
> 28.42
>
> 184.19
>
> 166.83
>
> 194.05
>
> 222.20
>
> 370.43
>
> 28.04
>
> 183.16
>
> 166.42
>
> 193.67
>
> 223.05
>
> 369.72
>
> 28.61
>
> 181.29
>
> 165.70
>
> 193.41
>
> 221.29
>
> 371.28
>
> 27.69
>
> 182.04
>
> 164.74
>
> 194.34
>
> 222.48
>
> 371.00
>
> 28.17
>
> 183.32
>
> 166.75
>
> -Xms1500m -Xmx1500m -Xem:server jnt.scimark2.commandline -large
>
> Composite Score
>
> FFT
>
> (1048576)
>
> SOR
>
> (1000*1000)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=100000,
>
> nz=1000000)
>
> LU
>
> (1000*1000)
>
> 179.31
>
> 37.93
>
> 359.34
>
> 27.18
>
> 289.51
>
> 182.60
>
> 178.31
>
> 35.84
>
> 359.34
>
> 28.08
>
> 288.78
>
> 179.50
>
> 179.35
>
> 37.19
>
> 258.66
>
> 28.08
>
> 289.43
>
> 183.40
>
> 179.02
>
> 35.63
>
> 360.01
>
> 27.14
>
> 289.92
>
> 182.40
>
> 179.80
>
> 37.44
>
> 360.01
>
> 27.25
>
> 290.08
>
> 184.21
>
>
> Sun sdk1.5
>
> -Xms1500m -Xmx1500m -server jnt.scimark2.commandline
>
> Composite Score
>
> FFT
>
> (1024)
>
> SOR
>
> (100*100)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=1000,nz=5000)
>
> LU
>
> (100*100)
>
> 427.30
>
> 252.57
>
> 593.82
>
> 22.51
>
> 321.41
>
> 946.18
>
> 431.48
>
> 272.11
>
> 596.21
>
> 22.16
>
> 322.68
>
> 944.21
>
> 432.80
>
> 273.99
>
> 596.77
>
> 22.54
>
> 322.20
>
> 948.48
>
> 428.75
>
> 256.96
>
> 596.03
>
> 22.58
>
> 323.63
>
> 944.54
>
> 432.90
>
> 276.25
>
> 597.32
>
> 22.59
>
> 323.16
>
> 945.19
>
>
> -Xms1500m -Xmx1500m –server jnt.scimark2.commandline -large
>
> Composite Score
>
> FFT
>
> (1048576)
>
> SOR
>
> (1000*1000)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=100000,
>
> nz=1000000)
>
> LU
>
> (1000*1000)
>
> 243.25
>
> 36.42
>
> 553.20
>
> 34.72
>
> 381.71
>
> 265.18
>
> 278.28
>
> 37.74
>
> 576.72
>
> 39.89
>
> 369.94
>
> 367.11
>
> 266.89
>
> 37.42
>
> 575.21
>
> 41.22
>
> 368.48
>
> 312.11
>
> 271.74
>
> 37.63
>
> 577.16
>
> 39.48
>
> 371.28
>
> 333.17
>
> 269.53
>
> 37.49
>
> 574.99
>
> 41.12
>
> 368.88
>
> 325.20
>
>
> gcj-4.0.2 –O3
>
> Composite Score
>
> FFT
>
> (1024)
>
> SOR
>
> (100*100)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=1000,
>
> nz=5000)
>
> LU
>
> (100*100)
>
> 214.69
>
> 228.30
>
> 360.18
>
> 11.19
>
> 151.84
>
> 321.94
>
> 220.42
>
> 195.46
>
> 338.18
>
> 7.96
>
> 276.17
>
> 284.33
>
> 254.33
>
> 214.59
>
> 360.18
>
> 11.58
>
> 277.23
>
> 408.05
>
> 179.55
>
> 184.54
>
> 355.71
>
> 6.71
>
> 143.22
>
> 227.56
>
> 233.90
>
> 215.02
>
> 360.58
>
> 11.57
>
> 276.41
>
> 305.92
>
> -large
>
> Composite Score
>
> FFT
>
> (1048576)
>
> SOR
>
> (1000*1000)
>
> Monte Carlo
>
> Sparse matmult
>
> (N=100000,
>
> nz=1000000)
>
> LU
>
> (1000*1000)
>
> 192.24
>
> 29.62
>
> 348.23
>
> 11.55
>
> 222.95
>
> 348.86
>
> 177.07
>
> 35.24
>
> 322.72
>
> 8.16
>
> 232.94
>
> 286.25
>
> 174.29
>
> 35.02
>
> 331.95
>
> 9.75
>
> 249.63
>
> 245.09
>
> 196.79
>
> 27.28
>
> 347.29
>
> 11.50
>
> 255.12
>
> 342.76
>
> 179.69
>
> 37.69
>
> 349.346
>
> 10.69
>
> 176.19
>
> 324.57
>
>
>
> --
> From : Simon.Chow@Software School of Fudan University
>