 |
Software > Rational >
|
 |
 |
 |
Performance information for the MASS libraries for Blue Gene/L and Blue Gene/P
|
| | | Abstract | | This document provides performance data for the MASS scalar and vector libraries for Blue Gene/L and Blue Gene/P, and comparison with the standard system math library. | | | | | | | | Content | | The following table provides approximate performance data for the MASS libraries running on Blue Gene/P. (The results are also representative of Blue Gene/L MASS.) The data was obtained by timing many repetitions of a length-1000 loop (for scalar functions) or calls for length-1000 vectors (for vector functions) over 1000 random vectors and includes all overheads. Timing in this way brings the input and output vectors into the on-chip cache (because the loop is short enough for the vectors to fit in the cache). Performance may deteriorate significantly when the input and output vectors are not in the cache. Performance may also deteriorate for arguments at or near the end-points of the valid argument ranges. The system library measurements were made with the versions of the library available on the test system. They may vary from the versions timed for previous versions of MASS. Users may experience performance that differs from that found in this table. The column labelled libm give the results from using the functions in the MASS C/C++ source code library libmassv.c to call the functions in the system math library. The C source code was compiled with the IBM Visual Age C/C++ compiler using the -O option. The columns labelled libmass and libmassv obtain the results using the libmass.a and libmassv.a library respectively. The columns labelled libmass speedup and libmassv speedup give the ratio of the corresponding entries from libm and libmass, libmassv respectively. Vector function names are used to indicate the corresponding scalar functions in libm and libmass. A blank entry in any column indicates that the corresponding function is not available. The range key at the end of this section describes the ranges from which random arguments were uniformly generated. MASS Performance (cycles per call, length 1000 vector/loop)
R a n g libm libmass libmass libmassv libmassv function e speedup speedup ---------------------------------------------------------- vacos B 205.66 172.18 1.19 41.73 4.93 vacosh G 477.03 203.22 2.35 45.62 10.46 vasin B 208.98 170.89 1.22 41.64 5.02 vasinh D 476.88 187.64 2.54 43.42 10.98 vatan B 207.60 120.11 1.73 D 231.65 147.85 1.57 vatan2 D 418.40 171.02 2.45 48.45 8.64 vatanh B 356.64 169.98 2.10 33.42 10.67 vcbrt D 267.13 125.91 2.12 17.88 14.94 vcopysign D 43.41 48.03 0.90 vcos B 152.31 71.03 2.14 29.24 5.21 D 246.61 101.43 2.43 29.24 8.43 vcosh D 261.29 96.04 2.72 18.66 14.00 vcosisin B 325.94 122.08 2.67 63.10 5.17 D 519.34 164.14 3.16 69.08 7.52 vdint D 8.51 vdiv D 3.36 vdnint D 64.97 73.06 0.89 7.68 8.46 verf C 98.23 65.64 1.50 verfc C 192.62 137.89 1.40 vexp D 200.50 90.03 2.23 14.77 13.57 vexpm1 D 192.66 95.14 2.03 16.10 11.97 vhypot D 283.92 108.07 2.63 vlgamma H 493.77 280.90 1.76 vlog C 329.11 120.14 2.74 16.04 20.52 vlog10 C 391.41 114.08 3.43 15.38 25.45 vlog1p H 209.71 115.99 1.81 22.92 9.15 vpow C 651.42 198.01 3.29 67.96 9.59 vqdrt C 22.92 vrcbrt D 17.08 vrec D 2.51 vrqdrt C 23.09 vrsqrt C 214.18 60.11 3.56 5.08 42.16 vsacos B 240.16 108.94 2.20 30.00 8.01 vsacosh G 337.93 156.26 2.16 32.77 10.31 vsasin B 220.88 109.52 2.02 29.38 7.52 vsasinh D 313.06 164.63 1.90 35.16 8.90 vsatan B 144.32 84.12 1.72 D 151.50 129.57 1.17 vsatan2 D 285.82 145.80 1.96 57.15 5.00 vsatanh B 311.06 145.14 2.14 30.63 10.16 vscbrt D 250.94 101.96 2.46 14.05 17.86 vscopysign D 43.44 46.24 0.94 vscos B 129.74 61.18 2.12 24.53 5.29 D 240.74 96.08 2.51 24.38 9.87 vscosh D 294.08 84.22 3.49 28.47 10.33 vscosisin B 60.42 D 67.09 vsdiv D 9.86 vserf C 99.38 54.13 1.84 vserfc C 183.33 70.14 2.61 vsexp D 242.95 83.28 2.92 14.05 17.29 vsexpm1 D 181.54 77.62 2.34 13.86 13.10 vshypot D 241.18 109.94 2.19 vsin B 173.32 66.23 2.62 29.21 5.93 D 273.44 104.90 2.61 29.21 9.36 vsincos B 317.60 94.21 3.37 50.31 6.31 D 511.76 135.25 3.78 50.32 10.17 vsinh D 264.30 101.14 2.61 27.18 9.72 vslgamma H 312.57 270.11 1.16 vslog C 200.60 88.05 2.28 13.82 14.52 vslog10 C 248.32 88.04 2.82 11.66 21.30 vslog1p H 200.78 97.76 2.05 16.84 11.92 vspow C 664.22 126.30 5.26 38.31 17.34 vsqdrt C 15.34 vsqrt C 182.18 64.94 2.81 8.49 21.46 vsrcbrt D 14.22 vsrec D 4.57 vsrint D 51.28 52.11 0.98 vsrqdrt C 15.29 vsrsqrt C 5.75 vssin B 112.71 69.09 1.63 24.11 4.67 D 238.69 95.39 2.50 24.10 9.90 vssincos B 44.46 D 44.63 vssinh D 285.09 85.14 3.35 25.31 11.26 vssqrt C 7.01 vstan D 292.65 129.82 2.25 31.25 9.36 vstanh F 329.74 79.97 4.12 27.65 11.93 vtan D 369.43 114.11 3.24 43.64 8.47 vtanh F 347.13 110.31 3.15 24.53 14.15 Range Key A 0,1 B -1,1 C 0,100 D -100,100 E -10,10 F -20,20 G 1,100 H -1,101 I 0,10 | | | | | | | Original publication date | | 2005/4/7 | | | | | |
 |
| IBM, the IBM logo and ibm.com are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml. |
 |
 |
 |
| Please take a moment to complete this form to help us better serve you. |
 |
 |
 |
|
|
|
 |
 |
| Product categories: |
 |
| | Software |  |
| | Software Development |  |
| | Traditional Progamming Language & Compilers |  |
| | Mathematical Acceleration Subsystem |  |
| | Libraries |  |
 |
| Operating system(s): |
| |
Linux
|
 |
| Software version: |
| |
4.4
|
 |
| Reference #: |
| |
7006977
|
 |
| IBM Group: |
| | Software Group |
 |
| Modified date: |
| | 2007-11-11 |
 |
|