Changes between Version 1 and Version 2 of Nofib


Ignore:
Timestamp:
Apr 2, 2010 4:17:35 PM (5 years ago)
Author:
batterseapower
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • Nofib

    v1 v2  
     1Nofib is the standard Haskell benchmark suite. It is described in the paper [http://www.haskell.org/ghc/docs/papers/nofib.ps.gz The nofib Benchmark Suite of Haskell Programs].
     2
     3== Getting nofib ==
     4
     5From your GHC tree, run:
     6
     7{{{
     8./darc-all --nofib get
     9}}}
     10
     11It will be pulled into the a "nofib" subdirectory.
    112
    213== Benchmarking an optimisation ==
     
    718
    819{{{
    9 make -k >& log1
    10 make -k EXTRA_HC_OPTS=-fenable-cool-optimisation >& log2
    11 nofib-analyse log1 log2
     20cd nofib
     21make clean && make boot && make -k >& log1
     22make clean && make boot && make -k EXTRA_HC_OPTS=-fenable-cool-optimisation >& log2
     23nofib-analyse/nofib-analyse log1 log2
    1224}}}
    1325
    14 The output of the nofib-analyse tool is quite readable. Be warned: missing values in the output typically mean that the benchmark crashed and may indicate a problem with your optimisation.
     26The output of the nofib-analyse tool is quite readable, with two provisios:
     27  * Missing values in the output typically mean that the benchmark crashed and may indicate a problem with your optimisation
     28  * If a difference between the two modes is displayed as an absolute quantity instead of a percentage, it means that the difference was below the threshold at which the analyser considers it significant
    1529
    1630If the comparison identifies any particularly bad benchmark results, you can run them individually by changing into their directory and running something like: