search

XtractOpen / Meganet.jl

A fresh approach to deep learning written in Julia
http://www.xtract.ai/
Other
14 stars 9 forks source link

Remove views from sparseKernel.jl #23

Closed klensink closed 6 years ago

klensink commented 6 years ago

19 micro benchmarks for JthetaTmv were done on arrays that were too big. The use of views was not worth their overhead on 32x32 images and actually slowed down the code.

This PR reduces the runtime on 512 examples from 144s to 133s.

julia> judge(hist)
BenchmarkTools.TrialJudgement: 
  time:   -5.39% => improvement (5.00% tolerance)
  memory: -4.00% => improvement (1.00% tolerance)
coveralls commented 6 years ago

Coverage Status

Coverage increased (+0.08%) to 71.349% when pulling 3f09740e3c61d690a2135582e36c3cc77f605264 on fixsparseKernel into a0128248bccae6a40795ad3693b60075af84c901 on master.