diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2020-07-17 12:47:56 +0100 |
|---|---|---|
| committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2020-07-20 07:34:56 +0000 |
| commit | 9c82e014260a997fe784affc7e0545972c3511e5 (patch) | |
| tree | 404b1ab89ab9b62b9883c182f81b4bf4b53448ad /arm_compute/runtime/CL/CLTuner.h | |
| parent | ba2cc1aea6bcd16b3ad81b55be18911af83d2113 (diff) | |
| download | ComputeLibrary-9c82e014260a997fe784affc7e0545972c3511e5.tar.gz | |
COMPMID-3604: Graph failures during tuning
Update ICLTuner interface to account for the new memory injection
interface.
Redirect to appropriate kernel execution interface depending on if the
kernel supports memory injection or not.
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I8ce29f5c22f1865c9e688d12b65e68ee4486f99c
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3588
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'arm_compute/runtime/CL/CLTuner.h')
| -rw-r--r-- | arm_compute/runtime/CL/CLTuner.h | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/arm_compute/runtime/CL/CLTuner.h b/arm_compute/runtime/CL/CLTuner.h index 745d57a959..aa31181d2d 100644 --- a/arm_compute/runtime/CL/CLTuner.h +++ b/arm_compute/runtime/CL/CLTuner.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2019 Arm Limited. + * Copyright (c) 2017-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -116,6 +116,7 @@ public: // Inherited methods overridden: void tune_kernel_static(ICLKernel &kernel) override; void tune_kernel_dynamic(ICLKernel &kernel) override; + void tune_kernel_dynamic(ICLKernel &kernel, const InputTensorMap &inputs, const OutputTensorMap &outputs) override; /** Is the kernel_event set ? * @@ -130,7 +131,7 @@ private: * * @return The optimal LWS to use */ - cl::NDRange find_optimal_lws(ICLKernel &kernel); + cl::NDRange find_optimal_lws(ICLKernel &kernel, const InputTensorMap &inputs, const OutputTensorMap &outputs); std::unordered_map<std::string, cl::NDRange> _lws_table; cl::Event _kernel_event; |