- condense conditions by `LOG_IF`
- only log memory use once after all tops

[style] Clean Up Logging for Net Init

Solver Refactor: Separate files and Change Solver's Type to String

…he kernel

Now PReLU backward is taking the same time as forward

Code cleanup
Removed unnecessary code
Fixed indent
merge if(channed_shared_)

[build] Qualify messages issued by CMake when CUDA is unavailable

[docs] direct installation questions to caffe-users

[docs] mark cuDNN v3 compatible

[test] drop bogus OpenCV guard for layer type

PReLU speed up

[example] Fix class label index error in tutorial 00.

Set policies to NEW to silence warnings in CMake 3.02 and later.

Open LMDB files with MDB_NOLOCK if no write access

Clean includes

this code seems not to apply to the caffe head.
```
if diff.ndim != 4:
                raise Exception('{} diff is not 4-d'.format(top))
```

diff.ndim != 4 is outdated

Project does not compile without opencv_imgcodecs in the library path if
you're using OpenCV 3.

This introduces a OPENCV_VERSION flag in Makefile.config that includes
the library if set to 3.  (Trying to include it with OpenCV 2 also
breaks the build)

Add opencv_imgcodecs to library path in Makefile

Yet another batch normalization PR

Use six library to ensure pycaffe.py python3 compliance

…lob-float-no-register

[pycaffe] fix boost 1.60 compatibility issue; close BVLC#3494

This is temporary measure to avoid an apparent upstream issue with
protobuf 3.0.0b2.post1.

Travis: force protobuf 3.0.0b2 for Python 3

This provides a framework for automatically aligning different layers of
a net despite up/downsampling, padding, and output size rounding.

Removed lint script reference to non-existant caffe_memcpy function.

- document by docstring and comment
- pep8
- add latest layers and alphabetize
- respect default crop params
- handle graphs with compositions of crops by walking only the
  first, cropped bottom of Crop layers
- make python3 happy by replacing arg tuple unpacking

- crop -> offset
- adjust crop axis by 1

- test known mappings: conv-pool-deconv stack, ReLU and 1x1 conv
- test effects of padding
- test rectangular/anisotropic coordinate mapping, test N-D
- catch error cases: negative crop, scale mismatch, tops that are not
  spatially connected

configure offset(s) through proto definition.

Changes are:
  reduce test blob dims for speed
  use standard Gaussian filler,
  polish formatting and rename tests,
  test HW crop and 5D crop,
  standard gradient checks

ND Crop layer

Python/net spec coordinate map and crop offset computation

Refine P2PSync

Add functions to check and grab GPU