You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

Adam 67b59ddb38 Minor MKL-DNN conv int8 performance fixes (#20753) ... test=develop		6 years ago
..
details	…
dynload	Improve elementwise operators performance in same dimensions. (#19763)	6 years ago
CMakeLists.txt	make_conv_workspace_size_configurable, test=develop (#20662)	6 years ago
collective_helper.cc	supports multiple NCCL communicators preserved in NCCLCommContext (#19407)	6 years ago
collective_helper.h	supports multiple NCCL communicators preserved in NCCLCommContext (#19407)	6 years ago
cpu_helper.cc	…
cpu_helper.h	…
cpu_helper_test.cc	…
cpu_info.cc	move_flags_to_unified_files_for_management, test=develop (#19224)	6 years ago
cpu_info.h	…
cpu_info_test.cc	…
cuda_device_function.h	…
cuda_device_guard.cc	…
cuda_device_guard.h	…
cuda_helper.h	refine PADDLE_ENFORCE codes for unify PADDLE_ASSERT_MSG (#19603)	6 years ago
cuda_helper_test.cu	…
cuda_primitives.h	…
cuda_profiler.h	…
cudnn_desc.h	paddle::framework::vectorize() templatization [PART3] (#19643)	6 years ago
cudnn_desc_test.cc	…
cudnn_helper.h	fix pool2d pool3d,support asymmetric padding and channel_last (#19739)	6 years ago
cudnn_helper_test.cc	…
cudnn_workspace_helper.cc	make_conv_workspace_size_configurable, test=develop (#20662)	6 years ago
cudnn_workspace_helper.h	make_conv_workspace_size_configurable, test=develop (#20662)	6 years ago
device_code.cc	Integrate NVRTC to support compiling CUDA kernel at runtime (#19422)	6 years ago
device_code.h	Integrate NVRTC to support compiling CUDA kernel at runtime (#19422)	6 years ago
device_code_test.cc	Integrate NVRTC to support compiling CUDA kernel at runtime (#19422)	6 years ago
device_context.cc	fix cuda dev_ctx allocator cmake deps, test=develop (#19953)	6 years ago
device_context.h	Enable users to create custom cpp op outside framework. (#19256)	6 years ago
device_context_test.cu	…
device_memory_aligment.cc	Make fuse_optimizer_op_pass also work when the model contains sparse gradients. (#18664)	6 years ago
device_memory_aligment.h	Make fuse_optimizer_op_pass also work when the model contains sparse gradients. (#18664)	6 years ago
device_tracer.cc	…
device_tracer.h	…
enforce.cc	…
enforce.h	Paddle error message stack shaping and optimization (#19895)	6 years ago
enforce_test.cc	add support to gcc8, add docker env test=develop (#19807)	6 years ago
event.h	…
flags.cc	test=develop, add communicator_is_sgd_optimizer flag (#20677)	6 years ago
float16.h	…
float16_test.cc	replace part of PADDLE_ASSERT to PADDLE_ENFORCE (#19285)	6 years ago
float16_test.cu	replace part of PADDLE_ASSERT to PADDLE_ENFORCE (#19285)	6 years ago
for_range.h	…
gpu_info.cc	enable cpu machine to run paddle in gpu lib	6 years ago
gpu_info.h	GPU allocation uses fraction of available memory (#18896)	6 years ago
hostdevice.h	…
init.cc	Fix dgc nan by stripping nccl from sparseReduce. (#20630)	6 years ago
init.h	Fix dgc nan by stripping nccl from sparseReduce. (#20630)	6 years ago
init_test.cc	Add signal message to stderr (#19421)	6 years ago
lock_guard_ptr.h	…
lodtensor_printer.cc	…
lodtensor_printer.h	…
lodtensor_printer_test.cc	…
macros.h	…
mkldnn_helper.h	Minor MKL-DNN conv int8 performance fixes (#20753)	6 years ago
mkldnn_reuse.h	Revert "Refactor conv computeINT8" (#20640)	6 years ago
nccl_helper.h	refine PADDLE_ENFORCE codes for unify PADDLE_ASSERT_MSG (#19603)	6 years ago
ngraph_helper.h	…
place.cc	Clean unused code of dim and place (#18565)	6 years ago
place.h	Clean unused code of dim and place (#18565)	6 years ago
place_test.cc	Clean unused code of dim and place (#18565)	6 years ago
port.h	…
profiler.cc	…
profiler.cu	refine PADDLE_ENFORCE codes for unify PADDLE_ASSERT_MSG (#19603)	6 years ago
profiler.h	…
profiler.proto	…
profiler_test.cc	…
stream_callback_manager.cc	…
stream_callback_manager.h	…
timer.cc	…
timer.h	…
timer_test.cc	…
transform.h	…
transform_test.cu	…
variant.h	…