提交 · 052b2d928e1fc2b2c478b4e576ba215f7c00e80c · testgroup / pytensor

25 9月, 2012 4 次提交
- updated doc · 052b2d92
  由 Ian Goodfellow 提交于 9月 24, 2012
  
  052b2d92
- add some type checking · df5c67f8
  由 Ian Goodfellow 提交于 9月 24, 2012
  
  df5c67f8
- Merge pull request #945 from nouiz/gpu_reduce_shape0 · 2196f4a4
  由 goodfeli 提交于 9月 24, 2012
```
Gpu reduce shape0
```
  2196f4a4
- fix crash of gpusum gpumax when the input shape contain zeros that we sum over · da11f0f4
  由 Frederic 提交于 9月 24, 2012
  
  da11f0f4
24 9月, 2012 3 次提交
- small crash fix in new PR. · f97497fb
  由 Frederic 提交于 9月 24, 2012
  
  f97497fb
- Merge pull request #934 from goodfeli/gpu_max · 91c8fcf3
  由 nouiz 提交于 9月 24, 2012
```
Gpu max
```
  91c8fcf3
- fix hash and code version · c3d358c5
  由 Ian Goodfellow 提交于 9月 24, 2012
  
  c3d358c5
22 9月, 2012 9 次提交
- fixed unit test that didn't handle exceptions correctly · 371eb713
  由 Ian Goodfellow 提交于 9月 22, 2012
  
  371eb713
- commented perform method, since it's not implemented for all cases · a817341a
  由 Ian Goodfellow 提交于 9月 22, 2012
  
  a817341a
- fixed bug in optimization · eb09a9a2
  由 Ian Goodfellow 提交于 9月 21, 2012
  
  eb09a9a2
- Merge pull request #944 from nouiz/fix_new_test · a5965456
  由 goodfeli 提交于 9月 21, 2012
```
This fix a new test for a new feature.
```
  a5965456
- This fix a new test for a new feature. · e5cbb431
  由 Frederic Bastien 提交于 9月 21, 2012
```
If travis would have run the tests, it would have detected it, but it didn't ran...
```
  e5cbb431
- Merge pull request #943 from nouiz/fix_conv2d_crash · 7c8a4b62
  由 lamblin 提交于 9月 21, 2012
```
This fix a crash recently added in 5178e4
```
  7c8a4b62
- Merge pull request #940 from goodfeli/eval · 7d8bcfed
  由 nouiz 提交于 9月 21, 2012
```
added eval method and tests
```
  7d8bcfed
- Merge pull request #942 from nouiz/gpu_alloc2 · 18708214
  由 lamblin 提交于 9月 21, 2012
```
Gpu alloc2
```
  18708214
- This fix a crash recently added in 5178e4 · 26875b37
  由 Frederic Bastien 提交于 9月 21, 2012
```
We need to update the second var before we use it!
```
  26875b37
21 9月, 2012 18 次提交
- Update test following the postponing of allocation of the device structure of CudaNdarray. · a3987a12
  由 Frederic 提交于 9月 21, 2012
  
  a3987a12
- make sure it work with non scalar tensor and test it. · c056820d
  由 Frederic 提交于 9月 20, 2012
  
  c056820d
- more clear error msg. · a9e821f5
  由 Frederic 提交于 9月 20, 2012
  
  a9e821f5
- use the new apply_nodes attribute name. · 6873dc73
  由 Frederic 提交于 9月 20, 2012
  
  6873dc73
- postpone the creating of the device structure to when we need it. · f441295b
  由 Frederic 提交于 9月 20, 2012
```
This is a significant speed up with the gc as most of the time, we
don't need it and allocating on the GPU is slow.
```
  f441295b
- don't need to call the c fct when the data is not allocated. · c50a2db1
  由 Frederic 提交于 9月 20, 2012
  
  c50a2db1
- move fct and force them to be inlined. · 460e9561
  由 Frederic 提交于 9月 20, 2012
  
  460e9561
- compile cuda_ndarray.cu with -O3 flags. · d0e0274f
  由 Frederic 提交于 9月 20, 2012
  
  d0e0274f
- New optimization that will use cudaMemset for GpuAlloc when we init memory to 0. · 7cc1d895
  由 Frederic 提交于 9月 20, 2012
  
  7cc1d895
- changed language of cautionary note · a3d9f911
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  a3d9f911
- test_max tests now do tests with 0 shape · d3687dee
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  d3687dee
- made CAReduce check for reducing 0-length axis with max/min · 0899d978
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  0899d978
- test reduce with shape 0 · 1e24b90d
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  1e24b90d
- got rid of cuda-specific scalar C code methods · 0975c51f
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  0975c51f
- fixed C code cache version for GpuCAReduce · 283546f5
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  283546f5
- made GpuCAReduce use scalar_op.c_code instead of cuda_assign_reduce · e0cca017
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  e0cca017
- updated comment--Fred is going to handle uncaught case in a different PR · 9411f9e8
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  9411f9e8
- Merge pull request #939 from nouiz/c_code_gpusub · c8934e50
  由 lamblin 提交于 9月 20, 2012
```
C code gpusub
```
  c8934e50
20 9月, 2012 6 次提交
- added eval method to tutorial · 500db9d1
  由 Ian Goodfellow 提交于 9月 20, 2012
  
  500db9d1
- use the new attribute name to remove deprecation warning. · 02274d20
  由 Frederic 提交于 9月 20, 2012
  
  02274d20
- remove g++ warning about arythmetic on void* · 39401c9b
  由 Frederic 提交于 9月 20, 2012
  
  39401c9b
- Now GpuSubtensor have c code by reusing the Subtensor helper c code generator. · e6b53352
  由 Frederic 提交于 9月 19, 2012
  
  e6b53352
- Set the macro in the nvcc compiler for the new numpy api. · 529f4421
  由 Frederic 提交于 9月 19, 2012
  
  529f4421
- make pydotprint add the timming in the plot with the new profiler. · a2666307
  由 Frederic 提交于 9月 19, 2012
  
  a2666307