Commits · d57ef7d309ad9feeda8b3523bd601ba30daa9918 · submodule / ngraph

22 Feb, 2018 2 commits
- style fix · d57ef7d3
  pthoreho authored Feb 22, 2018
  
  d57ef7d3
- - elementwise Add mkldnn support in cpu emitter · 42b4f4d0
  pthoreho authored Feb 22, 2018
  
  42b4f4d0
21 Feb, 2018 1 commit
- Merge pull request #513 from NervanaSystems/jmenon/mkldnn-compile · 233e4b1b
  Jayaram Bobba authored Feb 21, 2018
```
MKLDNN Emitter
```
  233e4b1b
20 Feb, 2018 15 commits
- Merge branch 'master' into jmenon/mkldnn-compile · e05a356e
  Jayaram Bobba authored Feb 20, 2018
  
  e05a356e
- Some more reduction ops (forward prop) (#510) · 96cabff0
  Adam Procter authored Feb 20, 2018
```
* Add product op

* Add Max (max reduce) and Min (min reduce) ops

* Refactor arithmetic reduction ops to a common base class

* Fix PREFER_EIGEN codepaths in cpu_emitter
```
  96cabff0
- Merge branch 'master' into jmenon/mkldnn-compile · d8199edc
  Jayaram Bobba authored Feb 20, 2018
  
  d8199edc
- Fixed bug in initializing weights for bn fprop mkldnn emitted code (#519) · 292ab46f
  Pruthvi authored Feb 20, 2018
```
* fixed bn weights initialization to correct size

* style fix
```
  292ab46f
- CPU: Revert prefix change · fd869f1e
  Jaikrishnan Menon authored Feb 20, 2018
  
  fd869f1e
- Merge branch 'master' into mkldnn-compile · 5d3d8f53
  Jaikrishnan Menon authored Feb 20, 2018
  
  5d3d8f53
- Merge pull request #512 from NervanaSystems/pruthvi/max_pooling · 5280405e
  Jayaram Bobba authored Feb 20, 2018
```
mkldnn MaxPoolBprop implementation in CPU emitter
```
  5280405e
- Merge branch 'master' into pruthvi/max_pooling · 94fe470c
  Jayaram Bobba authored Feb 20, 2018
  
  94fe470c
- CPU: Implement workaround for GCC < 5 (#518) · 67c97f6a
  Jai Menon authored Feb 20, 2018
  
  67c97f6a
- Merge branch 'master' into mkldnn-compile · 4e1d1ac7
  Jaikrishnan Menon authored Feb 20, 2018
  
  4e1d1ac7
- Merge branch 'master' into pruthvi/max_pooling · cbec3ad8
  Pruthvi authored Feb 20, 2018
  
  cbec3ad8
- add mxnet sockeye Seq2Seq model (#508) · 607bcbc4
  Ashok Emani authored Feb 20, 2018
```
* add mxnet sockeye Seq2Seq model

* update test with sockeye model
```
  607bcbc4
- Addressed PR review comments · 67fb65b8
  pthoreho authored Feb 20, 2018
  
  67fb65b8
- Cosmetic change while we update the IAT design document so · b48b4122
  Jaikrishnan Menon authored Feb 20, 2018
```
unnecessary prefixes can be avoided
```
  b48b4122
- Update copyright header · 9da5e7c1
  Jaikrishnan Menon authored Feb 20, 2018
  
  9da5e7c1
17 Feb, 2018 8 commits
- CPU: Dilated convolution and refactor · 1404a20b
  Jaikrishnan Menon authored Feb 17, 2018
  
  1404a20b
- CPU: Formatting fixes · 42e52e42
  Jaikrishnan Menon authored Feb 17, 2018
  
  42e52e42
- Merge branch 'master' into mkldnn-compile · e0c3400b
  Jaikrishnan Menon authored Feb 17, 2018
  
  e0c3400b
- CPU: Remove old-style inline emission · fc5cda7c
  Jaikrishnan Menon authored Feb 17, 2018
  
  fc5cda7c
- CPU: Formatting fixes · 7fdd8d70
  Jaikrishnan Menon authored Feb 17, 2018
  
  7fdd8d70
- CPU: Propagate staged primitives to the runtime context · cd74b8f0
  Jaikrishnan Menon authored Feb 17, 2018
  
  cd74b8f0
- CPU: Add primitive vector getter · c9c01be1
  Jaikrishnan Menon authored Feb 17, 2018
  
  c9c01be1
- CPU: Invoke interface for memory primitives · 154daf06
  Jaikrishnan Menon authored Feb 17, 2018
  
  154daf06
16 Feb, 2018 7 commits
- style fix · 466854c6
  pthoreho authored Feb 16, 2018
  
  466854c6
- Merge remote-tracking branch 'origin/master' into pruthvi/max_pooling · 5ce09de0
  pthoreho authored Feb 16, 2018
  
  5ce09de0
- style fix · c6672b3d
  pthoreho authored Feb 16, 2018
  
  c6672b3d
- - Added test for max pooling to verify mkldnn maxpool implementation · 03f6c0ab
  pthoreho authored Feb 16, 2018
```
- added workaround to attach the maxpool workspace for bprop delta propogation
```
  03f6c0ab
- CPU: Memory primitive initialization · 51de579e
  Jaikrishnan Menon authored Feb 16, 2018
  
  51de579e
- Update functions to conform to naming convention (#503) · fbf5a0cf
  Robert Kimball authored Feb 16, 2018
```
* change Emit* function to templated
```
  fbf5a0cf
- CPU: Switch to primitive pointers · 01a3d6b3
  Jaikrishnan Menon authored Feb 16, 2018
  
  01a3d6b3
15 Feb, 2018 7 commits
- Merge pull request #504 from NervanaSystems/jmenon/targetopts · a77cbda0
  Jayaram Bobba authored Feb 15, 2018
```
Remove additional target attributes since we already supply
```
  a77cbda0
- Merge branch 'master' into jmenon/targetopts · cca10627
  Jai Menon authored Feb 15, 2018
  
  cca10627
- Remove coding style violations, greatly simplify (#505) · e567c891
  Scott Cyphers authored Feb 15, 2018
  
  e567c891
- Remove additional target attributes since we already supply · 416725d7
  Jaikrishnan Menon authored Feb 15, 2018
```
the host CPU name in TargetOptions
```
  416725d7
- added bprop maxpool mkldnn implmenetation in the cpu emitted code · c4db6126
  pthoreho authored Feb 15, 2018
```
 - workaround to find the workspace requirement for maxpool bprop based on the forward_op reference
```
  c4db6126
- Doc for the rest of the Cs. (#501) · ab1cb2e6
  Scott Cyphers authored Feb 15, 2018
```
* Doc for the rest of the Cs.

* Review comments.
```
  ab1cb2e6
- CPU: Conv builder · 165de8f2
  Jaikrishnan Menon authored Feb 15, 2018
  
  165de8f2