Scott McKay fc51473b09 Update BFCArena logic to use backoff if cudaMalloc fails. Makes behaviour equivalent to when a CPU allocation fails. Add unit test. (#2748) ... Clear error when throwing an exception for a failed CUDA call so that there is only one error mechanism being used at a time. Minor improvements to logging to aid debugging of BFCArena behaviour.		2020-01-22 14:21:21 +10:00
..
codegen	Simplify cache implementation and avoid static variables that may carry over between models	2019-12-20 21:04:17 -08:00
common	Add BatchParallelFor, TryParallelFor, TryBatchParallelFor into ThreadPool (#2476)	2019-11-27 00:32:26 -08:00
framework	Update BFCArena logic to use backoff if cudaMalloc fails. Makes behaviour equivalent to when a CPU allocation fails. Add unit test. (#2748)	2020-01-22 14:21:21 +10:00
graph	Add OneHotEncoder and HashOneHotEncoder kernels. (#2830)	2020-01-13 17:58:33 -08:00
language_interop_ops	Introduce PrimitiveType into a Type System along with an integer constant (#2307)	2019-11-08 17:47:06 -08:00
mlas	MLAS: enable threading for quantized GEMMs (#2844)	2020-01-15 19:25:40 -08:00
optimizer	Optimize BatchNormalization to NCHWc Conv (#2855)	2020-01-20 16:35:03 -08:00
platform	Add basic stacktrace output for posix debug builds. (#2749)	2020-01-03 13:43:57 +10:00
protobuf	Bump up ONNX to the latest commit (#1868)	2019-09-20 18:15:16 -07:00
providers	Update BFCArena logic to use backoff if cudaMalloc fails. Makes behaviour equivalent to when a CPU allocation fails. Add unit test. (#2748)	2020-01-22 14:21:21 +10:00
session	InferenceSession::Run needs to call OnRunEnd for any EP that OnRunStart was called for so they can cleanup. Currently it only calls OnRunEnd if the Status is OK. Due to this the CUDA EP will throw during shutdown as the per-thread information has not been cleaned up prior to the CUDA library shutting down. (#2881)	2020-01-22 12:17:52 +10:00
util	Update protobuf to 3.11.2 (#1928)	2019-12-27 18:28:18 -08:00