7
0%
master: 0%

Ran 20 Dec 2016 06:34PM UTC

Files 21

Run time 1s

Badge

Embed ▾

Committed 20 Dec 2016 06:01PM UTC coverage: 54.23% (+4.6%) from 49.628%

Job # PROSPECT_DEV_MODE=1 TEST_SUITE=pa-examples

Build Type

push

travis-ci

Committed by

DrTodd13

Commit Message

Change indexed_dims field of InputInfo from array of 0 and 1 to array of Bool.

Change getOrAddRangeCorrelation in the following ways:
   1) Changed its input to stop dropping SingularDimensions.  The reason for this is that
      we have a rule that if all the dimensions are specified with a mask then there is a
      correlation to the size of the input array.  If you strip singular dimensions then
      this looks like there are all masks and so you can match a mask in one dimension with
      singular dimensions in the others and this would not correlate to the overall array
      correlation.  Also, for its input, I canonicalize individual singular dimensions
      so that all singular dimensions will match all others.
   2) Take an accounting of the kinds of dimensions, range|mask|singular.
   3) Add a case for one masked dimension and the rest singular dimensions.  Here, the range
      correlation is equivalent to the mask's correlation.

With these changes, the 3 inner parfors of the third parfor in kmeans (the one that handles
the centroids update line) all fuse.

Run Details

5769 of 10638 relevant lines covered (54.23%)

2186.32 hits per line

IntelLabs / ParallelAccelerator.jl / 966 / 7
0%
master: 0%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 966.7 (PROSPECT_DEV_MODE=1 TEST_SUITE=pa-examples)

IntelLabs / ParallelAccelerator.jl / 966 / 7 0% master: 0%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 966.7 (PROSPECT_DEV_MODE=1 TEST_SUITE=pa-examples)

IntelLabs / ParallelAccelerator.jl / 966 / 7
0%
master: 0%

README BADGES
x