diff options
author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2017-07-05 20:05:23 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-09-17 14:15:39 +0100 |
commit | 7d323a6adca97c130a0fc7c6299c75d581906edd (patch) | |
tree | 666be55f734cf5493641f3d99ba0afd12cf362f7 /arm_compute/runtime | |
parent | ab0a77edcb9f48de2aad216323b791d0dd95a3cd (diff) | |
download | ComputeLibrary-7d323a6adca97c130a0fc7c6299c75d581906edd.tar.gz |
COMPMID-440, COMPMID-441 - Port CLConvolutionLayer and CLFullyConnectedLayer to support 16 bit fixed point
Change-Id: I8d8ef2cb5ec453eb83fba8d8077550b96ed4bceb
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79837
Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Diffstat (limited to 'arm_compute/runtime')
-rw-r--r-- | arm_compute/runtime/CL/functions/CLConvolutionLayer.h | 4 | ||||
-rw-r--r-- | arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h | 2 |
2 files changed, 3 insertions, 3 deletions
diff --git a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h index 50a7dc95eb..aba88bd856 100644 --- a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h @@ -53,7 +53,7 @@ public: CLConvolutionLayerReshapeWeights(); /** Set the input and output tensors. * - * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QS8/F16/F32. + * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QS8/QS16/F16/F32. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p weights. * @param[out] output Destination tensor. Data types supported: Same as @p weights. * @param[in] transpose1xW True if the weights are to undergo a 1xW transposition after reshaping (in case of GEMM operation), false otherwise. @@ -88,7 +88,7 @@ public: * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], * while every optional dimension from 4 and above represent a batch of inputs. - * Data types supported: QS8/F16/F32. + * Data types supported: QS8/QS16/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported:Same as @p input. * @param[out] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs. diff --git a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h index 807ff693bc..64d56894d3 100644 --- a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h +++ b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h @@ -50,7 +50,7 @@ public: CLFullyConnectedLayerReshapeWeights(); /** Set the input and output tensors. * - * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QS8/F16/F32. + * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QS8/QS16/F16/F32. * @param[out] output Destination tensor. Data type supported: Same as @p input. * @param[in] transpose_weights True if the weights must be transposed. Data types supported: Same as @p weights. * @param[in] is_batched_fc_layer True if it is a batched fully connected layer |