diff options
author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2017-07-04 16:46:32 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-09-17 14:15:39 +0100 |
commit | 2bbd96457e3740fd9df5556607514b5e80a25720 (patch) | |
tree | 679935dd849bdac044769dfff67516962493dd51 /arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h | |
parent | 8a383694445dfebb84732b19d5b3299961e8ffe3 (diff) | |
download | ComputeLibrary-2bbd96457e3740fd9df5556607514b5e80a25720.tar.gz |
COMPMID-436, COMPMID-437 - Port NEConvolutionLayer & NEFullyConnectedLayer to support 16 bit fixed point
Change-Id: I69edf2dac242f941bac95c8479d921e7be6abca7
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79725
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Pablo Tello <pablo.tello@arm.com>
Diffstat (limited to 'arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h')
-rw-r--r-- | arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h b/arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h index c0ecafcd39..1eed4e7a84 100644 --- a/arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h +++ b/arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h @@ -47,7 +47,7 @@ public: ~NEGEMMMatrixAccumulateBiasesKernel() = default; /** Set the accumulate buffer and the biases of the kernel. * - * @param[in, out] accum The accumulate tensor to convert. Data type supported: QS8/F32 + * @param[in, out] accum The accumulate tensor to convert. Data type supported: QS8/QS16/F32 * @param[in] biases The shared biases tensor to append. It must be 1D Tensor. Data type supported: Same as @p input */ void configure(ITensor *accum, const ITensor *biases); |