diff options
Diffstat (limited to 'src/cpu/kernels/CpuDirectConv3dKernel.h')
| -rw-r--r-- | src/cpu/kernels/CpuDirectConv3dKernel.h | 24 |
1 files changed, 11 insertions, 13 deletions
diff --git a/src/cpu/kernels/CpuDirectConv3dKernel.h b/src/cpu/kernels/CpuDirectConv3dKernel.h index c7dcb0fb5e..fc64e8518b 100644 --- a/src/cpu/kernels/CpuDirectConv3dKernel.h +++ b/src/cpu/kernels/CpuDirectConv3dKernel.h @@ -39,10 +39,7 @@ class CpuDirectConv3dKernel : public ICpuKernel public: CpuDirectConv3dKernel() = default; ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE(CpuDirectConv3dKernel); - /** Set the src, weights, and dst tensor info. - * - * Valid data layouts: - * - NDHWC + /** Set the src, weights, biases and dst tensor info. * * Valid data type configurations: * |src0 |src1 |src2 |dst | @@ -50,34 +47,35 @@ public: * |F16 |F16 |F16 |F16 | * |F32 |F32 |F32 |F32 | * - * @param[in, out] src Input tensor info. - * @param[in] weights Set of kernels to convolve the input volume. + * @param[in, out] src0 Input tensor info. + * @param[in] src1 Set of kernels to convolve the input volume. * The 2nd dimension must be the same as the input's volume 1st dimension. - * @param[in] biases Set of biases. Can be nullptr. + * @param[in] src2 Set of biases. Can be nullptr. * @param[out] dst Output tensor info. * The 1st dimensions must be equal to the 1st dimension of the @p kernels tensor. * @param[in] conv_info Contains padding, stride, acitvation information. * */ - void configure(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, ITensorInfo *dst, const Conv3dInfo &conv_info); + void configure(const ITensorInfo *src0, const ITensorInfo *src1, const ITensorInfo *src2, ITensorInfo *dst, const Conv3dInfo &conv_info); /** Static function to check if given info will lead to a valid configuration * * Similar to CpuDirectConv3dKernel::configure() * * @return a status */ - static Status validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, const Conv3dInfo &conv_info); + static Status validate(const ITensorInfo *src0, const ITensorInfo *src1, const ITensorInfo *src2, const ITensorInfo *dst, const Conv3dInfo &conv_info); // Inherited methods overridden: void run_op(ITensorPack &tensors, const Window &window, const ThreadInfo &info) override; const char *name() const override; private: - /* Template function for convolution NDHWC */ - template <typename T> - void convolve_ndhwc(const Window &window, const ITensor *src, const ITensor *weights, const ITensor *biases, ITensor *dst); + /* Template function for convolution 3d NDHWC */ + using DirectConv3dKernelPtr = std::add_pointer<void(const ITensor *, const ITensor *, const ITensor *, ITensor *, const Conv3dInfo &, const Window &)>::type; - Conv3dInfo _conv_info{}; + Conv3dInfo _conv_info{}; + DirectConv3dKernelPtr _run_method{ nullptr }; + std::string _name{}; }; } // namespace kernels } // namespace cpu |