[nightly] DPH Performance Test Succeeded
DPH Buildbot
dph at ouroborus.net
Fri Jun 24 17:51:07 CEST 2011
Full logs at http://log.ouroborus.net/limitingfactor/dph
Environment
Platform
host: limitingfactor.cse.unsw.EDU.AU
arch: i386
processor: i386
system: Darwin 10.7.0
Versions
GHC The Glorious Glasgow Haskell Compilation System, version 7.0.1
GCC i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Total tests = 708
-- WINNERS (had a swing of < -10%)
dph.quicksort.vectorised.par.N1
runtime (wall clock) : 600.7ms ( -14) / 671.0ms ( -4) / 701.3ms (----)
kernel runtime (wall clock) : 394.0ms ( -3) / 409.5ms ( -3) / 417.0ms ( -3)
dph.quicksort.vectorised.par.N8
runtime (wall clock) : 801.3ms (----) / 801.5ms ( -1) / 801.7ms ( -11)
kernel runtime (wall clock) : 466.0ms (----) / 481.8ms (----) / 495.0ms ( -1)
dph.smvm.vectorised.par.N4
runtime (wall clock) : 100.6ms (----) / 100.6ms (----) / 100.8ms (----)
kernel runtime (wall clock) : 9.000ms (----) / 12.20ms ( -19) / 18.00ms (----)
repa.laplace.par.N1
runtime (wall clock) : 5.704s ( -39) / 5.794s ( -38) / 5.805s ( -40)
kernel runtime (wall clock) : 5.166s ( -25) / 5.204s ( -25) / 5.257s ( -24)
repa.laplace.par.N2
runtime (wall clock) : 3.503s ( -33) / 3.593s ( -32) / 3.604s ( -33)
kernel runtime (wall clock) : 3.021s ( -22) / 3.121s ( -20) / 3.166s ( -20)
repa.laplace.par.N8
runtime (wall clock) : 2.202s ( -15) / 2.263s ( -15) / 2.303s ( -15)
kernel runtime (wall clock) : 1.849s ( -12) / 1.874s ( -12) / 1.899s ( -13)
repa.blur.par.N2
runtime (wall clock) : 700.7ms (----) / 700.9ms ( -1) / 701.3ms ( -12)
kernel runtime (wall clock) : 430.0ms ( -3) / 433.2ms ( -3) / 436.0ms ( -8)
repa.canny.par.N1
runtime (wall clock) : 400.6ms (----) / 400.7ms (----) / 400.9ms (----)
kernel runtime (wall clock) : 21.00ms ( -12) / 21.40ms ( -11) / 22.00ms ( -8)
repa.canny.par.N2
runtime (wall clock) : 300.6ms (----) / 300.7ms (----) / 300.8ms (----)
kernel runtime (wall clock) : 15.00ms ( -12) / 16.90ms ( -10) / 18.00ms ( -10)
repa.fft2d.par.N2
runtime (wall clock) : 5.004s ( -9) / 5.184s ( -7) / 5.404s ( -5)
kernel runtime (wall clock) : 4.397s ( -11) / 4.600s ( -8) / 4.766s ( -8)
noslow.dph-prim.seq.Double.filterin_r
kernel runtime (wall clock) : 36.75us ( -14) / 36.76us ( -14) / 36.78us ( -14)
-- LOSERS (had a swing of > +10%)
dph.dotp.vectorised.par.N4
runtime (wall clock) : 1.801s ( -31) / 1.802s ( -31) / 1.802s ( -33)
kernel runtime (wall clock) : 21.00ms (----) / 32.30ms ( +18) / 35.00ms ( +13)
dph.evens.vectorised.par.N8
runtime (wall clock) : 500.8ms ( +25) / 501.0ms ( +22) / 501.1ms (----)
kernel runtime (wall clock) : 78.00ms ( +15) / 79.70ms ( +11) / 82.00ms ( -5)
dph.quicksort.vectorised.par.N2
runtime (wall clock) : 600.9ms ( +20) / 601.0ms ( -9) / 601.2ms ( -14)
kernel runtime (wall clock) : 342.0ms ( +13) / 363.3ms ( -2) / 371.0ms ( -5)
dph.quicksort.vectorised.par.N4
runtime (wall clock) : 601.0ms ( +20) / 671.2ms ( +6) / 701.4ms (----)
kernel runtime (wall clock) : 308.0ms (----) / 359.2ms (----) / 377.0ms (----)
dph.quickhull.vectorised.seq.N4
runtime (wall clock) : 900.8ms ( +12) / 901.0ms ( +6) / 901.4ms (----)
kernel runtime (wall clock) : 513.0ms (----) / 525.3ms ( +1) / 528.0ms (----)
dph.quickhull.vector-mutable.seq.N4
runtime (wall clock) : 300.7ms (----) / 380.9ms ( +23) / 401.2ms (----)
kernel runtime (wall clock) : 53.00ms (----) / 53.90ms (----) / 56.00ms (----)
dph.quickhull.vector-forkIO.par.N8
runtime (wall clock) : 300.6ms ( -25) / 320.8ms ( -20) / 501.0ms ( +25)
kernel runtime (wall clock) : 41.00ms ( +8) / 45.80ms ( +15) / 50.00ms ( +22)
repa.mmult.par.N2
runtime (wall clock) : 4.802s ( +12) / 5.144s ( +18) / 8.207s ( +86)
kernel runtime (wall clock) : 4.492s ( +6) / 4.838s ( +14) / 7.904s ( +86)
repa.mmult.par.N4
runtime (wall clock) : 2.502s ( +32) / 2.564s ( +35) / 2.617s ( +38)
kernel runtime (wall clock) : 2.253s ( +26) / 2.263s ( +26) / 2.273s ( +27)
repa.mmult.par.N8
runtime (wall clock) : 1.407s ( +17) / 1.427s ( +17) / 1.523s ( +16)
kernel runtime (wall clock) : 1.134s ( +5) / 1.148s ( +5) / 1.199s ( +2)
repa.laplace.par.N4
runtime (wall clock) : 2.502s ( -17) / 2.963s ( -1) / 3.303s ( +10)
kernel runtime (wall clock) : 2.052s ( +2) / 2.543s ( +26) / 2.892s ( +43)
repa.canny.par.N6
runtime (wall clock) : 200.5ms (----) / 240.6ms ( +20) / 300.6ms ( +50)
kernel runtime (wall clock) : 13.00ms ( +8) / 13.00ms ( +3) / 13.00ms (----)
repa.fft3d.par.N2
runtime (wall clock) : 18.51s ( +9) / 19.44s ( +12) / 19.92s ( +13)
kernel runtime (wall clock) : 15.25s ( +10) / 16.16s ( +13) / 16.68s ( +14)
repa.fft3d.par.N4
runtime (wall clock) : 11.11s ( +39) / 11.28s ( +41) / 11.31s ( +41)
kernel runtime (wall clock) : 9.285s ( +50) / 9.394s ( +51) / 9.444s ( +52)
repa.fft3d.par.N6
runtime (wall clock) : 8.709s ( +9) / 8.809s ( +9) / 8.909s ( +9)
kernel runtime (wall clock) : 7.277s ( +10) / 7.384s ( +11) / 7.481s ( +11)
noslow.list.*Double.find_indices
kernel runtime (wall clock) : 436.0us ( +12) / 437.9us ( +13) / 439.0us ( +13)
noslow.vector.Primitive.*Double.filterin_r
kernel runtime (wall clock) : 407.8us ( +13) / 409.5us ( +13) / 410.6us ( +13)
noslow.vector-unsafe.Primitive.*Double.filterin_r
kernel runtime (wall clock) : 409.6us ( +13) / 411.3us ( +13) / 412.5us ( +13)
noslow.dph-prim.seq.Double.scale_r
kernel runtime (wall clock) : 69.05us ( +10) / 69.06us ( +10) / 69.08us ( +10)
noslow.dph-prim.seq.Double.axpy
kernel runtime (wall clock) : 91.30us ( +11) / 91.32us ( +11) / 91.34us ( +11)
noslow.vector.Primitive.Double.scale_r
kernel runtime (wall clock) : 69.28us ( +12) / 69.29us ( +12) / 69.30us ( +12)
noslow.vector.Primitive.Double.axpy
kernel runtime (wall clock) : 91.12us ( +11) / 91.14us ( +11) / 91.15us ( +11)
noslow.vector.Unboxed.Double.scale_r
kernel runtime (wall clock) : 69.23us ( +11) / 69.27us ( +11) / 69.31us ( +11)
noslow.vector-unsafe.Primitive.Double.scale_r
kernel runtime (wall clock) : 65.70us ( +11) / 65.71us ( +11) / 65.73us ( +11)
noslow.vector-unsafe.Unboxed.Double.scale_r
kernel runtime (wall clock) : 70.17us ( +12) / 70.18us ( +12) / 70.20us ( +12)
More information about the Cvs-ghc
mailing list