55 cl::NDRange fake_lws(2000);
Bifrost based OpenCL tuner implementation.
static CLScheduler & get()
Access the scheduler singleton.
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
void set_lws_hint(const cl::NDRange &lws_hint)
Set the Local-Workgroup-Size hint.
1 channel, 1 F32 per channel
ARM_COMPUTE_EXPECT(has_error==expected, framework::LogLevel::ERRORS)
Interface for the direct convolution kernel.
SimpleTensor< float > src
Copyright (c) 2017-2021 Arm Limited.
Interface to enqueue OpenCL kernels and get/set the OpenCL CommandQueue and ICLTuner.
DatasetMode
Possible dataset modes.
TEST_SUITE_END() FIXTURE_DATA_TEST_CASE(RunSmall
[CLActivationLayer Test snippet]
TEST_SUITE(U8_to_S8) FIXTURE_DATA_TEST_CASE(RunSmall
Padding and stride information class.
void tune_kernel_static(ICLKernel &kernel) override
Tune OpenCL kernel statically.
void set_target(GPUTarget target)
Set the targeted GPU architecture.
TEST_CASE(FusedActivation, framework::DatasetMode::ALL)
Validate fused activation expecting the following behaviours:
void default_init(ICLTuner *cl_tuner=nullptr, CLGEMMHeuristicsHandle *gemm_h=nullptr)
Initialises the context and command queue used by the scheduler to default values and sets a default ...
void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info)
Set the input, weights, biases and output tensors.