Commits · 88f5d9a3dc85f372d00c5f95d1dc5240c422233f · 钟尚武 / dlib

25 Oct, 2015 2 commits
- Implemented more cuDNN bindings and cleaned up code a bit. · 88f5d9a3
  Davis King authored Oct 25, 2015
  
  88f5d9a3
- Made the add_layer variadic constructors take tuples of parameters in addition · e9efffff
  Davis King authored Oct 25, 2015
```
to just regular lists of parameters.
```
  e9efffff
24 Oct, 2015 1 commit
- Added methods to inspect where the current data is at. · 1175a7cc
  Davis King authored Oct 24, 2015
  
  1175a7cc
22 Oct, 2015 2 commits
- minor cleanup · be944398
  Davis King authored Oct 21, 2015
  
  be944398
- Added tensor spec. · 99e948db
  Davis King authored Oct 21, 2015
  
  99e948db
21 Oct, 2015 4 commits
- Added missing const · 37d493e2
  Davis King authored Oct 21, 2015
  
  37d493e2
- Added missing const · beae2212
  Davis King authored Oct 21, 2015
  
  beae2212
- Improve messages generated by cmake · beee4953
  Davis King authored Oct 21, 2015
  
  beee4953
- Changed tensor layout from NHWC to NCHW since this layout is much faster for · c49631dd
  Davis King authored Oct 20, 2015
```
cuDNN.
```
  c49631dd
20 Oct, 2015 10 commits
- Added gpu_data spec and also optimized it a little bit. · 30c4ee54
  Davis King authored Oct 20, 2015
  
  30c4ee54
- Added comments · fe8b99f1
  Davis King authored Oct 20, 2015
  
  fe8b99f1
- clarified comments · 99ec66b8
  Davis King authored Oct 19, 2015
  
  99ec66b8
- setup cuDNN forward convolution · 6a1e50a4
  Davis King authored Oct 19, 2015
  
  6a1e50a4
- Hid the cuDNN context in a thread local variable so the user doesn't · 10480cb9
  Davis King authored Oct 19, 2015
```
need to deal with it.
```
  10480cb9
- Added some of the cuDNN conv calls. · d63c4682
  Davis King authored Oct 19, 2015
  
  d63c4682
- Changed compiler versions to ones that have thread_local. · 1022b5b9
  Davis King authored Oct 19, 2015
  
  1022b5b9
- Made the C++11 test check for thread_local support. · a47fd477
  Davis King authored Oct 19, 2015
  
  a47fd477
- Require the compiler to support the thread_local C++11 feature. · c8d33ea6
  Davis King authored Oct 19, 2015
  
  c8d33ea6
- fixed comment · ad03f747
  Davis King authored Oct 19, 2015
  
  ad03f747
18 Oct, 2015 21 commits
- fixed uninitialized variable. · 7972f663
  Davis King authored Oct 18, 2015
  
  7972f663
- simplified code a little · 1a868800
  Davis King authored Oct 18, 2015
  
  1a868800
- merged · dac02110
  Davis King authored Oct 18, 2015
  
  dac02110
- Improved C++11 check · 0800fd26
  Davis King authored Oct 18, 2015
  
  0800fd26
- Renamed dlib-shared to dlib_shared because the - was causing nvcc to get · 22d1cd22
  Davis King authored Oct 18, 2015
```
confused when that name appeared on the command line.
```
  22d1cd22
- Made the cmake test for C++11 support try to compile a C++11 project if · 2c639d3a
  Davis King authored Oct 18, 2015
```
we don't know what compiler we are dealing with.  If the test compile works
then we assume we have C++11.
```
  2c639d3a
- merged · 7c991b18
  Davis King authored Oct 18, 2015
  
  7c991b18
- Made more move constructors use swap() just to be safe · 1569d590
  Davis King authored Oct 18, 2015
  
  1569d590
- Fixed some warnings and errors from visual studio 2015 · c1da9dc9
  Davis King authored Oct 18, 2015
  
  c1da9dc9
- Added another required C++11 feature. · 78109ac9
  Davis King authored Oct 18, 2015
  
  78109ac9
- Tweaked move constructor to the form I will probably use. · 1de8eaf8
  Davis King authored Oct 18, 2015
  
  1de8eaf8
- Fixed tensor move constructor · 5162c984
  Davis King authored Oct 18, 2015
  
  5162c984
- fixed move constructor · de34309c
  Davis King authored Oct 18, 2015
  
  de34309c
- Added top level build folder to the ignore list. · c9fd6d0b
  Davis King authored Oct 18, 2015
  
  c9fd6d0b
- Apparently you need gcc 4.7 or newer for this stuff. · 3a65e1c7
  Davis King authored Oct 18, 2015
  
  3a65e1c7
- minor cleanup · bf93f4f3
  Davis King authored Oct 18, 2015
  
  bf93f4f3
- Made CMake try to compile a CUDA test project to see if the compiler the user is trying · c5f320e6
  Davis King authored Oct 18, 2015
```
really supports the version of CUDA we found.
```
  c5f320e6
- optimized code a little · 681abe78
  Davis King authored Oct 18, 2015
  
  681abe78
- Moved gpu_data into its own file, fixed a few bugs, and cleaned up · 3ba21975
  Davis King authored Oct 18, 2015
```
a few things.
```
  3ba21975
- Fully implemented the gpu_data object and also cleaned up a few other minor · 0508fe2b
  Davis King authored Oct 18, 2015
```
details.
```
  0508fe2b
- Fixed cmake script so nvcc doesn't freak out. · 1e623983
  Davis King authored Oct 17, 2015
  
  1e623983