arm_compute v18.05
diff --git a/src/runtime/CL/functions/CLFullyConnectedLayer.cpp b/src/runtime/CL/functions/CLFullyConnectedLayer.cpp
index 2b4670b..151fa1b 100644
--- a/src/runtime/CL/functions/CLFullyConnectedLayer.cpp
+++ b/src/runtime/CL/functions/CLFullyConnectedLayer.cpp
@@ -37,10 +37,8 @@
namespace
{
-Status validate_mm(const ITensorInfo &input, const ITensorInfo &weights, const ITensorInfo &output, bool is_interleaved_transposed)
+Status validate_mm(const ITensorInfo &input, const ITensorInfo &weights, const ITensorInfo &output)
{
- const GPUTarget gpu_target = CLScheduler::get().target();
-
if(is_data_type_quantized_asymmetric(input.data_type()))
{
// Since we need negative offsets for computing convolution, we need to change QuantizationInfo()
@@ -55,7 +53,7 @@
}
else
{
- ARM_COMPUTE_RETURN_ON_ERROR(CLGEMMMatrixMultiplyKernel::validate(&input, &weights, &output, 1.f, is_interleaved_transposed, GEMMReshapeInfo(), gpu_target));
+ ARM_COMPUTE_RETURN_ON_ERROR(CLGEMM::validate(&input, &weights, nullptr, &output, 1.f, 0.0f, GEMMInfo(false, false, true /* Reshape weights only for the first run */)));
}
return Status{};
@@ -75,12 +73,12 @@
}
CLFullyConnectedLayer::CLFullyConnectedLayer(std::shared_ptr<IMemoryManager> memory_manager)
- : _memory_group(memory_manager), _im2col_kernel(), _reshape_weights_kernel(), _mm_kernel(), _mm_gemmlowp(memory_manager), _gemmlowp_output_stage(), _accumulate_biases_kernel(), _im2col_output(),
- _gemmlowp_output(), _reshape_weights_output(), _are_weights_reshaped(true), _is_fc_after_conv(true), _accumulate_biases(false), _is_quantized(false)
+ : _memory_group(memory_manager), _im2col_kernel(), _reshape_weights_kernel(), _mm_gemm(memory_manager), _mm_gemmlowp(memory_manager), _gemmlowp_output_stage(), _accumulate_biases_kernel(),
+ _im2col_output(), _gemmlowp_output(), _reshape_weights_output(), _are_weights_reshaped(true), _is_fc_after_conv(true), _accumulate_biases(false), _is_quantized(false), _original_weights(nullptr)
{
}
-void CLFullyConnectedLayer::configure_mm(const ICLTensor *input, const ICLTensor *weights, ICLTensor *output, bool is_interleaved_transposed)
+void CLFullyConnectedLayer::configure_mm(const ICLTensor *input, const ICLTensor *weights, ICLTensor *output)
{
if(_is_quantized)
{
@@ -102,8 +100,7 @@
else
{
// Configure matrix multiply kernel
- _mm_kernel.set_target(CLScheduler::get().target());
- _mm_kernel.configure(input, weights, output, 1.f, is_interleaved_transposed);
+ _mm_gemm.configure(input, weights, nullptr, output, 1.f, 0.0f, GEMMInfo(false, false, true /* Reshape weights only for the first run */));
}
}
@@ -114,7 +111,7 @@
// If the fully connected layer is called after a convolution layer, the input tensor must be linearized
// Initialize output tensor for im2col
- TensorShape shape_im2col = compute_im2col_shape(input->info());
+ TensorShape shape_im2col = compute_im2col_fc_shape(input->info());
_im2col_output.allocator()->init(input->info()->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(shape_im2col));
// Configure im2col kernel
@@ -122,7 +119,7 @@
_im2col_kernel.configure(input, &_im2col_output, Size2D(1, 1), PadStrideInfo(1, 1, 0, 0), false);
// Configure matrix multiply kernel
- configure_mm(&_im2col_output, weights, output, false);
+ configure_mm(&_im2col_output, weights, output);
// Allocate the output tensor for im2col once all the configure methods have been called
_im2col_output.allocator()->allocate();
@@ -133,7 +130,7 @@
ARM_COMPUTE_ERROR_ON(input->info()->dimension(0) != weights->info()->dimension(1));
// Configure matrix multiply kernel
- configure_mm(input, weights, output, false);
+ configure_mm(input, weights, output);
}
void CLFullyConnectedLayer::configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, bool transpose_weights, bool are_weights_reshaped)
@@ -152,6 +149,7 @@
_is_fc_after_conv = true;
_accumulate_biases = false;
_is_quantized = is_data_type_quantized_asymmetric(input->info()->data_type());
+ _original_weights = weights;
// Configure gemmlowp output
if(_is_quantized)
@@ -222,13 +220,6 @@
_gemmlowp_output_stage.configure(&_gemmlowp_output, biases, output, output_multiplier, output_shift, output->info()->quantization_info().offset);
_gemmlowp_output.allocator()->allocate();
}
-
- // Allocate the transpose tensor if the are_weights_reshaped flag is false and once all the configure methods have been called
- if(!_are_weights_reshaped)
- {
- // Allocate the tensor for the weights reshaped
- _reshape_weights_output.allocator()->allocate();
- }
}
Status CLFullyConnectedLayer::validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, bool transpose_weights, bool are_weights_reshaped)
@@ -243,7 +234,7 @@
bool is_quantized = is_data_type_quantized_asymmetric(input->data_type());
const GPUTarget gpu_target = CLScheduler::get().target();
- const ITensorInfo &im2col_input = TensorInfo(input->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(compute_im2col_shape(input)));
+ const ITensorInfo &im2col_input = TensorInfo(input->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(compute_im2col_fc_shape(input)));
const ITensorInfo &reshaped_weights = TensorInfo(weights->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(compute_transposed_shape(*weights)));
const ITensorInfo &gemmlowp_output = TensorInfo(output->clone()->set_is_resizable(true).reset_padding().set_data_type(DataType::S32));
@@ -300,7 +291,7 @@
ARM_COMPUTE_RETURN_ERROR_ON(input->dimension(0) != weights_to_use->dimension(1));
}
// Validate matrix multiply kernel
- ARM_COMPUTE_RETURN_ON_ERROR(validate_mm(*input_to_use, *weights_to_use, *tmp_output, false));
+ ARM_COMPUTE_RETURN_ON_ERROR(validate_mm(*input_to_use, *weights_to_use, *tmp_output));
// Validate output stage for asymmetric quantized types
if(is_quantized)
@@ -313,12 +304,7 @@
void CLFullyConnectedLayer::run()
{
- // Reshape of the weights (happens only once)
- if(!_are_weights_reshaped)
- {
- _are_weights_reshaped = true;
- _reshape_weights_kernel.run();
- }
+ prepare();
_memory_group.acquire();
@@ -335,7 +321,7 @@
}
else
{
- CLScheduler::get().enqueue(_mm_kernel, !_accumulate_biases);
+ _mm_gemm.run();
}
// Accumulate biases if provided
@@ -353,3 +339,30 @@
_memory_group.release();
}
+
+void CLFullyConnectedLayer::prepare()
+{
+ // Reshape of the weights (happens only once)
+ if(!_are_weights_reshaped)
+ {
+ ARM_COMPUTE_ERROR_ON(!_original_weights->is_used());
+
+ // Run reshape weights kernel and mark weights as unused
+ _reshape_weights_output.allocator()->allocate();
+ _reshape_weights_kernel.run();
+ _original_weights->mark_as_unused();
+
+ // Prepare GEMM prepare and release unused weights
+ if(!_is_quantized)
+ {
+ _mm_gemm.prepare();
+ if(!_reshape_weights_output.is_used())
+ {
+ _reshape_weights_output.allocator()->free();
+ }
+ }
+
+ CLScheduler::get().queue().finish();
+ _are_weights_reshaped = true;
+ }
+}