Changes between Version 95 and Version 96 of DataParallel/BenchmarkStatus

Show
Ignore:
Timestamp:
12/06/10 23:59:43 (2 years ago)
Author:
benl
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • DataParallel/BenchmarkStatus

    v95 v96  
    102102  || dph.sumsq.c.seq || 116ms || 3.48 || || || 
    103103 
    104   '''Status''': ok, but run again with LLVM to see if that fixes the slowdown wrt C[[br]] 
     104  '''Status''': fine[[br]] 
    105105 
    106106[[br]] 
    … …  
    119119  B: The vectorised.par version runs faster than vectorised.seq because the latter has a duplicate counter in the inner loop. We need a duplicate-loop-counter removal optimisation. 
    120120 
    121   '''Status''': ok, but need other versions for comparison[[br]] 
    122   '''Todo''': Check again with LLVM. 
     121  '''Status''': fine[[br]] 
    123122 
    124123[[br]] 
    … …  
    136135  A : Benchmark is totally memory bound, so we're not expecting to see much speedup. [[br]]  
    137136    
    138   '''Status''': ok, but need other versions for comparison. 
     137  '''Status''': ok, but run again with LLVM to see if that fixes the slowdown wrt C. 
    139138 [[br]] 
    140139 [http://darcs.haskell.org/packages/dph/examples/smvm/ SMVM] '''(BROKEN)'''::