Changes between Version 103 and Version 104 of DataParallel/BenchmarkStatus


Ignore:
Timestamp:
Dec 13, 2010 2:20:17 AM (4 years ago)
Author:
benl
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • DataParallel/BenchmarkStatus

    v103 v104  
    139139 
    140140 [[br]] 
    141  [http://darcs.haskell.org/packages/dph/examples/smvm/ SMVM] '''(BROKEN - Fusion)''':: 
     141 [http://darcs.haskell.org/packages/dph/examples/smvm/ SMVM] '''(BROKEN)''':: 
    142142 Multiplies a dense vector with a sparse matrix represented in the ''compressed sparse row format (CSR).'' [[br]] 
    143143 
     
    155155 
    156156[[br]] 
    157  [http://darcs.haskell.org/libraries/dph/dph-examples/spectral/QuickSort/ QuickSort] '''(BROKEN - SpecConstr) (SLOWDOWN)''':: 
     157 [http://darcs.haskell.org/libraries/dph/dph-examples/spectral/QuickSort/ QuickSort] '''(BROKEN) (SLOWDOWN)''':: 
    158158  Sort a vector of doubles by recursively splitting it and sorting the two halves. This is a naive benchmark used for regression testing only. We divide right down to two-point vectors and construct the result using copying append. A production algorithm would switch to an in-place sort once the size of the vector reaches a few thousand elements. N=100k. 
    159159 
     
    163163  || dph.quicksort.vectorised.par.N4 || 422ms ||  1.01 || || || 
    164164 
    165   '''Status''': Sequential vectorised version does not compile due to a blowup in !SpecConstr. 
     165  '''Status''': Sequential vectorised version does not compile due to a loop in !SpecConstr (#4831). 
    166166 
    167167[[br]] 
     
    191191These programs also use user defined algebraic data types. Vectorization of these programs is still a work in progress. 
    192192 
    193  [http://darcs.haskell.org/libraries/dph/dph-examples/imaginary/Words/ Words] '''(BROKEN - SpecConstr)''':: 
     193 [http://darcs.haskell.org/libraries/dph/dph-examples/imaginary/Words/ Words] '''(BROKEN)''':: 
    194194  Counts the number of words in a string. This is a naive divide-and-conquer benchmark that divides right down to a single character. A production program would switch to a simple sequential algorithm once the string chunks were small enough. It's a good stress test for the vectoriser though. 
    195195 
    196  '''Status''': Sequential vectorised version does not compile due to blowup in SpecConstr.[[br]] 
     196 '''Status''': Sequential vectorised version does not compile due to loop in !SpecConstr (#4831). LLVM versions take >10 min to compile (#4838)[[br]] 
    197197 '''Todo''': Generate some larger test data. Right now it's just got a small test string baked into the program. 
    198198 
    199199[[br]] 
    200  [http://darcs.haskell.org/libraries/dph/dph-examples/real/NBody/ BarnesHut] '''(BROKEN -fllvm)''' '''(SLOWLORIS)''':: 
     200 [http://darcs.haskell.org/libraries/dph/dph-examples/real/NBody/ BarnesHut] '''(BROKEN)''' '''(SLOWLORIS)''':: 
    201201  This benchmark implements the Barnes-Hut algorithm to solve the ''n''-body problem in two dimensions. There is a naive O(n^2^) version in the same package. 
    202202 
     
    207207  A : Time stated is end-to-end, not just for the kernel. 
    208208 
    209   '''Status''':  -fasm vesions compile but fusion doesn't work so it's very slow. LLVM versions take 30min to compile. [[br]] 
     209  '''Status''':  -fasm vesions compile but fusion doesn't work so it's very slow. LLVM versions take 30min to compile (#4838) [[br]] 
    210210  '''!ToDo''': Make the vectorised version give the same output as the vector version. The benchmark setup is a bit different. Fixing this won't cause a 50x speed difference though. 
    211211