2
70%
master: 70%

Ran 26 Nov 2016 04:12PM UTC

Files 81

Run time 3s

Badge

Embed ▾

Committed 26 Nov 2016 04:02PM UTC coverage: 72.347% (+0.04%) from 72.307%

Job # USE_SSE=ON USE_AVX=ON USE_DOUBLE=OFF

Build Type

push

travis-ci

Committed by

edgarriba

Commit Message

Subtle speed optimization (#419)

* test/test_core.h
  change type of variables cl_platform and cl_device from size_t to int to reduce warnings

tiny_dnn/layers/convolutional_layer.h
  change the way to set data to vector container

tiny_dnn/layers/layer.h
  reuse local variable diff to suppress heap memory allocation/deallocation (maybe futile...)
  only parallelize if size of data to process is big enough

tiny_dnn/layers/slice_layer.h
  remove supposedly garbled characters

tiny_dnn/optimizers/optimizer.h
  to control parallel_for.h `for_i` function, add 3rd argument `bool parallelize` to optimizer::update method

tiny_dnn/util/image.h
  remove implicit downcast warning by using holy C style casts

* optimize further by tuning up iterative call of activation function.

* add comment

* revert change in `test/test_core.h`

Coverage Stats

4029 of 5569 relevant lines covered (72.35%)

2747305.16 hits per line

tiny-dnn / tiny-dnn / 1043 / 2
70%
master: 70%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 1043.2 (USE_SSE=ON USE_AVX=ON USE_DOUBLE=OFF)

tiny-dnn / tiny-dnn / 1043 / 2 70% master: 70%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 1043.2 (USE_SSE=ON USE_AVX=ON USE_DOUBLE=OFF)

tiny-dnn / tiny-dnn / 1043 / 2
70%
master: 70%

README BADGES
x