[nightly] DPH Performance Test Succeeded

DPH Buildbot dph at ouroborus.net
Fri Feb 18 09:31:52 CET 2011


Full logs at http://log.ouroborus.net/limitingfactor/dph

Environment
  Platform
    host:      limitingfactor.cse.unsw.EDU.AU
    arch:      i386
    processor: i386
    system:    Darwin 10.6.0
  Versions
    GHC The Glorious Glasgow Haskell Compilation System, version 7.1.20110217
    GCC i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5659)

 Total tests = 702
 
 -- WINNERS (had a swing of < -10%)
 dph.sumsq.vectorised.par.N1
       runtime        (wall clock)    : 400.6ms ( -20) / 400.7ms ( -20) / 400.8ms ( -20)
       kernel runtime (wall clock)    : 368.0ms ( -17) / 368.1ms ( -17) / 369.0ms ( -17)
 
 dph.sumsq.vectorised.par.N2
       runtime        (wall clock)    : 300.5ms (----) / 300.6ms (----) / 300.8ms (----)
       kernel runtime (wall clock)    : 184.0ms ( -17) / 184.0ms ( -17) / 184.0ms ( -17)
 
 dph.sumsq.vectorised.par.N4
       runtime        (wall clock)    : 200.5ms (----) / 200.6ms (----) / 200.9ms (----)
       kernel runtime (wall clock)    : 92.00ms ( -16) / 92.00ms ( -17) / 92.00ms ( -17)
 
 dph.sumsq.vectorised.par.N8
       runtime        (wall clock)    : 100.5ms (----) / 100.5ms (----) / 100.6ms (----)
       kernel runtime (wall clock)    : 48.00ms ( -20) / 53.70ms ( -16) / 56.00ms ( -15)
 
 dph.sumsq.vectorised.seq.N4
       runtime        (wall clock)    : 400.6ms ( -20) / 400.7ms ( -20) / 400.8ms ( -20)
       kernel runtime (wall clock)    : 367.0ms ( -15) / 367.9ms ( -15) / 368.0ms ( -15)
 
 dph.sumsq.vector.seq.N4
       runtime        (wall clock)    : 400.6ms ( -20) / 400.7ms ( -20) / 400.9ms ( -20)
       kernel runtime (wall clock)    : 368.0ms ( -10) / 368.0ms ( -11) / 368.0ms ( -11)
 
 dph.evens.vectorised.par.N8
       runtime        (wall clock)    : 500.7ms (----) / 500.9ms (----) / 501.1ms (----)
       kernel runtime (wall clock)    : 70.00ms (  -7) / 71.50ms (  -8) / 73.00ms ( -21)
 
 dph.quicksort.vectorised.par.N2
       runtime        (wall clock)    : 601.1ms ( -14) / 671.3ms (  -4) / 701.8ms (----)
       kernel runtime (wall clock)    : 352.0ms (  -7) / 374.1ms (  -6) / 385.0ms (  -6)
 
 dph.quicksort.vectorised.par.N4
       runtime        (wall clock)    : 601.1ms ( -14) / 651.2ms (  -7) / 701.4ms (----)
       kernel runtime (wall clock)    : 332.0ms ( -14) / 361.4ms ( -11) / 371.0ms ( -12)
 
 dph.quicksort.vectorised.par.N8
       runtime        (wall clock)    : 801.3ms ( -11) / 801.5ms ( -12) / 801.6ms ( -20)
       kernel runtime (wall clock)    : 461.0ms ( -13) / 479.2ms ( -12) / 491.0ms ( -21)
 
 dph.quickhull.vectorised.par.N1
       runtime        (wall clock)    : 1.201s  ( -20) / 1.201s  ( -20) / 1.202s  ( -20)
       kernel runtime (wall clock)    : 725.0ms ( -32) / 726.6ms ( -32) / 729.0ms ( -32)
 
 dph.quickhull.vectorised.par.N2
       runtime        (wall clock)    : 900.9ms ( -10) / 921.3ms ( -12) / 1.001s  (  -9)
       kernel runtime (wall clock)    : 479.0ms ( -26) / 510.6ms ( -26) / 545.0ms ( -25)
 
 dph.quickhull.vectorised.par.N4
       runtime        (wall clock)    : 801.1ms (----) / 801.3ms ( -10) / 801.5ms ( -11)
       kernel runtime (wall clock)    : 366.0ms ( -30) / 390.0ms ( -32) / 407.0ms ( -34)
 
 dph.quickhull.vectorised.par.N8
       runtime        (wall clock)    : 801.3ms ( -11) / 802.3ms ( -11) / 804.3ms ( -12)
       kernel runtime (wall clock)    : 379.0ms ( -31) / 390.9ms ( -33) / 416.0ms ( -31)
 
 dph.quickhull.vectorised.seq.N4
       runtime        (wall clock)    : 1.001s  (  -9) / 1.001s  (  -9) / 1.001s  (  -9)
       kernel runtime (wall clock)    : 580.0ms ( -16) / 586.2ms ( -16) / 590.0ms ( -16)
 
 dph.quickhull.vector-mutable.seq.N4
       runtime        (wall clock)    : 400.7ms ( -20) / 400.9ms ( -20) / 401.0ms ( -20)
       kernel runtime (wall clock)    : 50.00ms ( -41) / 52.40ms ( -39) / 58.00ms ( -33)
 
 dph.quickhull.vector-forkIO.par.N4
       runtime        (wall clock)    : 400.6ms (----) / 400.8ms (----) / 401.1ms (----)
       kernel runtime (wall clock)    : 32.00ms ( -43) / 36.10ms ( -40) / 43.00ms ( -31)
 
 dph.quickhull.vector-forkIO.par.N8
       runtime        (wall clock)    : 400.8ms (----) / 400.9ms (----) / 401.0ms (----)
       kernel runtime (wall clock)    : 35.00ms ( -37) / 40.70ms ( -36) / 49.00ms ( -33)
 
 repa.laplace.par.N1
       runtime        (wall clock)    : 5.803s  ( -38) / 5.804s  ( -38) / 5.805s  ( -40)
       kernel runtime (wall clock)    : 3.372s  ( -51) / 3.375s  ( -51) / 3.381s  ( -51)
 
 repa.laplace.par.N2
       runtime        (wall clock)    : 3.302s  ( -37) / 3.473s  ( -34) / 3.503s  ( -35)
       kernel runtime (wall clock)    : 2.024s  ( -48) / 2.088s  ( -47) / 2.102s  ( -47)
 
 repa.laplace.par.N6
       runtime        (wall clock)    : 2.403s  ( -14) / 2.413s  ( -14) / 2.503s  ( -11)
       kernel runtime (wall clock)    : 1.742s  ( -19) / 1.784s  ( -17) / 1.833s  ( -17)
 
 repa.laplace.par.N8
       runtime        (wall clock)    : 2.103s  ( -19) / 2.104s  ( -21) / 2.105s  ( -22)
       kernel runtime (wall clock)    : 1.549s  ( -27) / 1.566s  ( -26) / 1.576s  ( -28)
 
 repa.blur.par.N1
       runtime        (wall clock)    : 900.9ms ( -69) / 951.0ms ( -67) / 1.001s  ( -68)
       kernel runtime (wall clock)    : 619.0ms ( -76) / 626.7ms ( -76) / 634.0ms ( -76)
 
 repa.blur.par.N2
       runtime        (wall clock)    : 600.7ms ( -62) / 600.9ms ( -62) / 601.1ms ( -62)
       kernel runtime (wall clock)    : 344.0ms ( -75) / 344.7ms ( -75) / 345.0ms ( -75)
 
 repa.blur.par.N4
       runtime        (wall clock)    : 400.6ms ( -56) / 440.7ms ( -51) / 500.8ms ( -44)
       kernel runtime (wall clock)    : 200.0ms ( -72) / 215.8ms ( -70) / 227.0ms ( -69)
 
 repa.blur.par.N6
       runtime        (wall clock)    : 400.6ms ( -43) / 400.8ms ( -43) / 401.0ms ( -43)
       kernel runtime (wall clock)    : 163.0ms ( -67) / 165.7ms ( -66) / 171.0ms ( -66)
 
 repa.blur.par.N8
       runtime        (wall clock)    : 300.6ms ( -50) / 330.8ms ( -45) / 400.7ms ( -34)
       kernel runtime (wall clock)    : 134.0ms ( -67) / 137.7ms ( -67) / 143.0ms ( -66)
 
 repa.edgedetect.par.N1
       runtime        (wall clock)    : 400.5ms ( -20) / 400.6ms ( -20) / 400.6ms ( -20)
       kernel runtime (wall clock)    : 48.00ms ( -76) / 48.70ms ( -76) / 50.00ms ( -76)
 
 repa.edgedetect.par.N2
       runtime        (wall clock)    : 300.5ms (----) / 300.6ms ( -23) / 300.7ms ( -25)
       kernel runtime (wall clock)    : 29.00ms ( -77) / 29.00ms ( -78) / 29.00ms ( -79)
 
 repa.edgedetect.par.N4
       runtime        (wall clock)    : 300.5ms (----) / 300.6ms (----) / 300.8ms (----)
       kernel runtime (wall clock)    : 18.00ms ( -76) / 22.00ms ( -73) / 24.00ms ( -73)
 
 repa.edgedetect.par.N6
       runtime        (wall clock)    : 300.6ms (----) / 300.6ms (----) / 300.8ms (----)
       kernel runtime (wall clock)    : 18.00ms ( -67) / 18.90ms ( -69) / 19.00ms ( -71)
 
 repa.edgedetect.par.N8
       runtime        (wall clock)    : 200.5ms ( -33) / 210.6ms ( -30) / 300.6ms (----)
       kernel runtime (wall clock)    : 16.00ms ( -68) / 18.00ms ( -67) / 20.00ms ( -68)
 
 repa.fft2d.par.N2
       runtime        (wall clock)    : 4.504s  ( -18) / 4.664s  ( -16) / 5.004s  ( -12)
       kernel runtime (wall clock)    : 4.071s  ( -17) / 4.175s  ( -17) / 4.473s  ( -13)
 
 repa.fft2d.par.N4
       runtime        (wall clock)    : 3.003s  ( -19) / 3.073s  ( -17) / 3.104s  ( -16)
       kernel runtime (wall clock)    : 2.695s  ( -18) / 2.721s  ( -18) / 2.743s  ( -17)
 
 repa.fft2d.par.N6
       runtime        (wall clock)    : 2.503s  ( -36) / 2.593s  ( -34) / 2.603s  ( -35)
       kernel runtime (wall clock)    : 2.237s  ( -38) / 2.262s  ( -38) / 2.278s  ( -38)
 
 repa.fft2d.par.N8
       runtime        (wall clock)    : 2.404s  ( -43) / 2.445s  ( -43) / 2.505s  ( -42)
       kernel runtime (wall clock)    : 2.153s  ( -45) / 2.184s  ( -45) / 2.266s  ( -43)
 
 repa.fft3d.par.N8
       runtime        (wall clock)    : 6.517s  ( -12) / 6.587s  ( -12) / 6.717s  ( -11)
       kernel runtime (wall clock)    : 5.559s  ( -10) / 5.621s  ( -10) / 5.772s  (  -9)
 
 noslow.dph-prim.seq.Double.axpy
       kernel runtime (wall clock)    : 82.32us ( -10) / 82.33us ( -10) / 82.35us ( -10)
 
 noslow.vector.Primitive.Double.scale_r
       kernel runtime (wall clock)    : 61.52us ( -11) / 61.53us ( -11) / 61.54us ( -11)
 
 noslow.vector.Primitive.Double.axpy
       kernel runtime (wall clock)    : 82.11us ( -10) / 82.12us ( -10) / 82.14us ( -10)
 
 -- LOSERS  (had a swing of > +10%)
 dph.dotp.vectorised.par.N1
       runtime        (wall clock)    : 2.502s  ( +14) / 2.502s  ( +14) / 2.503s  ( +14)
       kernel runtime (wall clock)    : 48.00ms ( -13) / 48.00ms ( -14) / 48.00ms ( -14)
 
 dph.dotp.vectorised.par.N2
       runtime        (wall clock)    : 2.502s  ( +14) / 2.552s  ( +16) / 2.602s  ( +18)
       kernel runtime (wall clock)    : 24.00ms ( -11) / 34.60ms (----) / 43.00ms (  -2)
 
 dph.dotp.vectorised.par.N4
       runtime        (wall clock)    : 2.602s  ( +13) / 2.602s  ( +13) / 2.603s  ( +13)
       kernel runtime (wall clock)    : 21.00ms (  -5) / 26.40ms (  -4) / 31.00ms (----)
 
 dph.dotp.vectorised.seq.N4
       runtime        (wall clock)    : 2.502s  (  +9) / 2.522s  ( +10) / 2.603s  ( +13)
       kernel runtime (wall clock)    : 53.00ms (  -7) / 53.00ms (  -9) / 53.00ms ( -15)
 
 dph.dotp.vector.seq.N4
       runtime        (wall clock)    : 2.502s  (  +9) / 2.562s  ( +11) / 2.603s  ( +13)
       kernel runtime (wall clock)    : 53.00ms (  -9) / 53.30ms ( -11) / 55.00ms ( -14)
 
 dph.evens.vectorised.par.N1
       runtime        (wall clock)    : 500.7ms (----) / 500.9ms (----) / 501.1ms (----)
       kernel runtime (wall clock)    : 205.0ms ( +13) / 205.3ms ( +12) / 206.0ms ( +12)
 
 dph.evens.vectorised.par.N2
       runtime        (wall clock)    : 500.8ms ( +25) / 500.9ms ( +25) / 501.0ms ( +25)
       kernel runtime (wall clock)    : 114.0ms ( +13) / 119.0ms ( +11) / 124.0ms ( +12)
 
 dph.evens.vectorised.par.N4
       runtime        (wall clock)    : 500.8ms ( +25) / 500.9ms ( +25) / 501.1ms ( +25)
       kernel runtime (wall clock)    : 83.00ms ( +15) / 86.70ms (  +9) / 89.00ms (  +5)
 
 dph.evens.vectorised.seq.N4
       runtime        (wall clock)    : 600.7ms ( +20) / 600.9ms ( +20) / 601.2ms ( +20)
       kernel runtime (wall clock)    : 191.0ms ( +19) / 191.2ms ( +10) / 192.0ms (----)
 
 dph.evens.vector.seq.N4
       runtime        (wall clock)    : 400.9ms (----) / 490.8ms ( +22) / 501.1ms ( +25)
       kernel runtime (wall clock)    : 92.00ms (  -5) / 92.60ms (  -5) / 93.00ms (  -5)
 
 repa.mmult.par.N1
       runtime        (wall clock)    : 16.51s  ( +92) / 16.51s  ( +91) / 16.52s  ( +88)
       kernel runtime (wall clock)    : 16.21s  ( +92) / 16.26s  ( +92) / 16.31s  ( +92)
 
 repa.mmult.par.N2
       runtime        (wall clock)    : 8.305s  ( +93) / 8.447s  ( +94) / 8.709s  ( +98)
       kernel runtime (wall clock)    : 8.080s  ( +91) / 8.214s  ( +93) / 8.414s  ( +98)
 
 repa.mmult.par.N4
       runtime        (wall clock)    : 4.302s  (+126) / 4.305s  (+126) / 4.305s  (+126)
       kernel runtime (wall clock)    : 4.042s  (+125) / 4.055s  (+126) / 4.064s  (+127)
 
 repa.mmult.par.N8
       runtime        (wall clock)    : 2.305s  ( +92) / 2.326s  ( +90) / 2.403s  ( +82)
       kernel runtime (wall clock)    : 2.029s  ( +87) / 2.067s  ( +88) / 2.113s  ( +80)
 
 repa.fft3d.par.N4
       runtime        (wall clock)    : 10.01s  ( +25) / 10.05s  ( +26) / 10.11s  ( +26)
       kernel runtime (wall clock)    : 8.448s  ( +36) / 8.542s  ( +38) / 8.619s  ( +39)
 
 noslow.list.*Double.count_filter
       kernel runtime (wall clock)    : 241.9us ( +11) / 242.8us ( +11) / 243.5us ( +12)
 
 noslow.vector-unsafe.boxed.*Double.filterout
       kernel runtime (wall clock)    : 378.4us ( +12) / 379.4us ( +12) / 380.2us ( +12)
 
 noslow.dph-prim.seq.Double.filterin_r
       kernel runtime (wall clock)    : 48.37us ( +32) / 48.45us ( +32) / 48.51us ( +32)
 
 noslow.vector.Unboxed.Double.filterin_r
       kernel runtime (wall clock)    : 46.84us ( +35) / 46.87us ( +35) / 46.90us ( +35)





More information about the Cvs-ghc mailing list