Changes between Version 1 and Version 2 of DataParallel/BenchmarkStatus


Ignore:
Timestamp:
May 1, 2008 7:15:17 AM (7 years ago)
Author:
gckeller
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • DataParallel/BenchmarkStatus

    v1 v2  
    33== Status of DPH Banchmarks == 
    44 
     5|| '''Program'''  || '''Sequential (manually vectorised) ''' || '''Vectorised'''         || ''' Parallel'''                     || 
     6|| DotP           ||Order of mag. faster than list impl      || Same performance as seq. || speedup of 2 for 2 CPUs, 4 threads  || 
     7|| !QuickSort     ||Slower than list (fusion)                || Slower than seq. (why?)  || speedup of 1.4 on 2 CPUs            || 
     8|| !SparseVector  ||Similar to DotP                          ||                          ||                                     || 
     9|| Primes (Nesl)  ||15 x faster than list version            || NYI                      || 20 x slower than seq (fusion?)      || 
     10|| Primes (Simon) ||NYI                                      || Working                  || NYI                                 || 
     11|| !BarnesHut     ||Small bug in alg                         || Working                  || See seq.                            ||   
     12  
     13 
     14General remarks: 
     15 
     16* I only ran a first set of benchmarks when checking for what's there. I'll run the benchmarks  
     17  properly as next step 
     18 
     19* Fusion doesn't work well on parallel programs yet, so for all but simple examples, the parallel program performs worse than the 
     20  sequential 
     21 
     22* The compiler doesn't explooit all fusion opportunities for QSort and BarnesHut. Once this is fixed, they should run considerably faster. 
     23 
     24 
     25 
     26 
     27