39 Commits (cd10cedee22c5df118db4aece71fd6e7700a5ab0)

Author	SHA1	Message	Date
chengduoZH	76e188e5c6	Add layer norm [GPU]	9 years ago
qingqing01	19367389c0	Update the CUDA kernel.	9 years ago
qingqing01	32d881beab	Optimize the rowwise add function.	9 years ago
QI JUN	efd3726929	remove unused place (#6972) ... * remove unused place * fix ci	9 years ago
dzhwinter	0d2235aadf	GPUPlace to CUDAPlace (#6960)	9 years ago
QI JUN	93a2d9c59d	add more place test and rename Cudnn to CUDNN (#6621) ... * add more place_test and rename Cudnn to CUDNN * fix ci	9 years ago
dzhwinter	0e9b393b34	"derived cudnnDevice context" (#6585) ... * "derived cudnnDevice context" * "leave remove cudnn handle from CUDADeviceContext" * "fix math function error"	9 years ago
QI JUN	61ec0b9516	Refine device context (#6433) ... There are mainly following fixes: - take `DeviceContext` as the template parameter of math functors and OpKernel instead of `Place` - remove `eigen_device` interface in base class `DeviceContext` - remove `GetEigenDevice` interface in `ExecutionContext` and base class `DeviceContext` - remove unused `platform::EigenDeviceConverter` - rename `REGISTER_OP_GPU_KERNEL` to `REGISTER_OP_CUDA_KERNEL` - rename `USE_GPU_ONLY_OP` to `USE_CUDA_ONLY_OP`	9 years ago
dangqingqing	7fb1f7a25f	Fix lstm_op and gru_op in debug mode.	9 years ago
Yang Yang(Tony)	18f0c40a97	feature/while_grad_op (#5554) ... * first commit * Python API for while op * Python Unittest for simple while_op forward * fix out to be list * Fix UT * VarType * Fix several bugs * Fix bug * Fix bug * Fix Bug * Fix bug * Fix unittest * Remove debug log * Add comments * add PADDLE_ENFORCE * while_grad_op first commit * Add `BlockDescBind::FindRecursiveOrCreateVar()` and fix bugs * not sure how to setdim of while outputs * push for test * add executor vlog * fix bug of while_op cond * Several enhancement for code 1. Backward always infer shape & infer var type. Since there are RENAME variables will be created when creating backward operator, but their shape & var types are not inferenced. 2. Never use SomePtr-> directly, since every pointer could be nullptr if it is a function return value. Add `detail::Ref` to cast pointer to reference safely. 3. Enhance error message for backward. 4. Infer data type of variable in `sum` and `tensor_write` * Fix bugs of while_op gradient * Fix several bugs of while_op grad * fix fill zeros like * fix 3 >= 3 * fix place holder shouldn't be null * fail on sum op * Fix SumOp of TensorList * clean up * pass while test * fix test_array_write_read * pass sum op * Support int/int64 for fill_constant_batch_size_like * Fix compile	9 years ago
dangqingqing	2673657684	Move RowwiseAdd functor to math_funcion and Add ColwiseSum functor.	9 years ago
dangqingqing	e9082bb78e	Resume unit testing.	9 years ago
dangqingqing	524ccba4fe	Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into cmake_speed	9 years ago
dangqingqing	f5e367655e	Use G++ to compile some cu operators.	9 years ago
emailweixu	58b4c9af34	Fixing duplicate struct name TensorSetConstant. (#5532) ... TensorSetConstant struct is used both in math_function.cc and math_function.cu. Somehow the release version can correctly handle it. But in debug version, set_constant_with_place() in math_function.cu uses the TensorSetConstant in math_function.cc and causes crash.	9 years ago
Yu Yang	0708a1550c	Fix CI	9 years ago
Yu Yang	aadb098138	Add `op::math::set_constant` without template	9 years ago
dangqingqing	cd38286684	Add gradient check unit testing and fix bug.	9 years ago
Markus Kliegl	164898277c	MatMul operator (#4856) ... * initial matmul operator Similar to np.matmul, but also has transpose_X and transpose_Y flags, and only supports tensors from rank 1 to 3 inclusive. For GPU, uses cublas?gemmStridedBatched. For CPU, uses cblas_?gemm_batch if available via MKL; otherwise a simple serial implementation that loops over the batch dimension is employed for now.	9 years ago
qijun	ab5dc9fe18	remove SelectedRows functors to selected_rows_functor.h	9 years ago
qijun	df2d1769fd	fix code style	9 years ago
qijun	7ef568e893	fix gpu unittest error	9 years ago
qijun	f59a7c1d36	add gpu functor for SelectedRows	9 years ago
guosheng	9ffa79cd91	Add gemm with stride	9 years ago
Yu Yang	81d56ca86b	Remove lazy-initialization in device_context ... * Also use `const DeviceContext&` all the time, to prevent `const_cast` Fix #4169 Fix #3468 Fix #3475	9 years ago
qijun	960a525550	fix gpu build error	9 years ago
qijun	2ec8dab4c7	follow comments	9 years ago
qijun	37aa4b98ff	refine unittest	9 years ago
qijun	c2631ebf6f	add unittest	9 years ago
qijun	331b304a18	merge baidu/develop	9 years ago
qijun	52b52ba80c	fix gpu build error	9 years ago
qijun	de967fcefe	set gemm support continuous memory now	9 years ago
qijun	7eb07b3399	fix conflict	9 years ago
qijun	8de4e3bdd6	disable gpu implementation temporarily	9 years ago
qijun	7307b439e1	fix gpu build error	9 years ago
qijun	5703eb50fa	add .clang-format file	9 years ago
qijun	97d8175a5e	add global matmul function for Tensor	9 years ago
qijun	f190a79538	fix gpu build error	9 years ago
qijun	22dac40c3a	add gemm for both cpu and gpu	9 years ago