diff options
| author | SiCong Li <sicong.li@arm.com> | 2022-01-28 18:24:39 +0000 |
|---|---|---|
| committer | SiCong Li <sicong.li@arm.com> | 2022-05-06 15:01:45 +0000 |
| commit | b63b1196adea8b07dd8db77c2492a212650deba0 (patch) | |
| tree | b264035197873f56c69784bec68cad7041b5d423 /src/runtime/CL | |
| parent | 3bb72b69566f18ad5c9446d318d2fc2b5f6dba42 (diff) | |
| download | ComputeLibrary-b63b1196adea8b07dd8db77c2492a212650deba0.tar.gz | |
Integrate Dynamic Fusion patches
* Add public interfaces:
* OperatorGraph: Describe a workload that could contain fused kernels
* IWorkload: Generic interface for workloads built from OperatorGraph
* ClWorkload: OpenCL workloads built from OperatorGraph
* ClCompositeOperator: Runtime async operator to execute a ClWorkload
* DependencyGraph (will likely be deprecated in later iterations)
* Add example
* cl_fused_conv2d_elementwise_add.cpp to explain how to use the new
interfaces
* Add internal translation layer
* Refactor ClKernelBuildingAPI
* Remove non-tile based gemm native kernel component
* Minor interface changes
* Add integration tests
Resolves COMPMID-5161
Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: Ib987ed79289ab0bcbd3130d54f5793408d9f1240
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7510
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/runtime/CL')
| -rw-r--r-- | src/runtime/CL/CLScheduler.cpp | 4 | ||||
| -rw-r--r-- | src/runtime/CL/CLTuner.cpp | 6 |
2 files changed, 5 insertions, 5 deletions
diff --git a/src/runtime/CL/CLScheduler.cpp b/src/runtime/CL/CLScheduler.cpp index 4cff707f1a..26124ed7e9 100644 --- a/src/runtime/CL/CLScheduler.cpp +++ b/src/runtime/CL/CLScheduler.cpp @@ -191,7 +191,7 @@ void CLScheduler::enqueue_common(ICLKernel &kernel, ITensorPack &tensors, bool f #if defined(ENABLE_EXPERIMENTAL_DYNAMIC_FUSION) -void CLScheduler::enqueue_common(ICLKernel &kernel, experimental::dynamic_fusion::TensorBinding &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc, bool flush) +void CLScheduler::enqueue_common(ICLKernel &kernel, ITensorPack &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc, bool flush) { ARM_COMPUTE_ERROR_ON_MSG(!_is_initialised, "The CLScheduler is not initialised yet! Please call the CLScheduler::get().default_init(), \ @@ -246,7 +246,7 @@ void CLScheduler::enqueue_op(ICLKernel &kernel, ITensorPack &tensors, bool flush #if defined(ENABLE_EXPERIMENTAL_DYNAMIC_FUSION) -void CLScheduler::enqueue_op(ICLKernel &kernel, experimental::dynamic_fusion::TensorBinding &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc, bool flush) +void CLScheduler::enqueue_op(ICLKernel &kernel, ITensorPack &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc, bool flush) { enqueue_common(kernel, tensors, exec_desc, flush); } diff --git a/src/runtime/CL/CLTuner.cpp b/src/runtime/CL/CLTuner.cpp index 81fe7dbde6..8ce5177847 100644 --- a/src/runtime/CL/CLTuner.cpp +++ b/src/runtime/CL/CLTuner.cpp @@ -68,7 +68,7 @@ private: #if defined(ENABLE_EXPERIMENTAL_DYNAMIC_FUSION) struct CompositeKernelData : public CLTuner::IKernelData { - CompositeKernelData(experimental::dynamic_fusion::TensorBinding &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc) + CompositeKernelData(ITensorPack &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc) : _tensors{ tensors }, _exec_desc{ exec_desc } { } @@ -80,7 +80,7 @@ struct CompositeKernelData : public CLTuner::IKernelData } private: - experimental::dynamic_fusion::TensorBinding &_tensors; + ITensorPack &_tensors; const experimental::dynamic_fusion::ClExecutionDescriptor &_exec_desc; }; #endif // defined(ENABLE_EXPERIMENTAL_DYNAMIC_FUSION) @@ -166,7 +166,7 @@ void CLTuner::tune_kernel_dynamic(ICLKernel &kernel, ITensorPack &tensors) } #if defined(ENABLE_EXPERIMENTAL_DYNAMIC_FUSION) -void CLTuner::tune_kernel_dynamic(ICLKernel &kernel, experimental::dynamic_fusion::TensorBinding &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc) +void CLTuner::tune_kernel_dynamic(ICLKernel &kernel, ITensorPack &tensors, const experimental::dynamic_fusion::ClExecutionDescriptor &exec_desc) { CompositeKernelData data{ tensors, exec_desc }; |