COMPMID-1722 : CL: Implement Range
Change-Id: I88da6eb5289c303b1dc91606c1560ce629746058
Reviewed-on: https://review.mlplatform.org/381
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/arm_compute/core/CL/CLHelpers.h b/arm_compute/core/CL/CLHelpers.h
index a86870a..78427c3 100644
--- a/arm_compute/core/CL/CLHelpers.h
+++ b/arm_compute/core/CL/CLHelpers.h
@@ -135,5 +135,14 @@
* @return True if the configuration is supported
*/
bool cl_winograd_convolution_layer_supported(const Size2D &output_tile, const Size2D &kernel_size, DataLayout data_layout);
+
+/** Helper function to get the preferred native vector width size for built-in scalar types that can be put into vectors
+ *
+ * @param[in] device A CL device
+ * @param[in] dt data type
+ *
+ * @return preferred vector width
+ */
+size_t preferred_vector_width(const cl::Device &device, DataType dt);
}
#endif /* __ARM_COMPUTE_CLHELPERS_H__ */
diff --git a/arm_compute/core/CL/CLKernels.h b/arm_compute/core/CL/CLKernels.h
index d89426d..cfcfb74 100644
--- a/arm_compute/core/CL/CLKernels.h
+++ b/arm_compute/core/CL/CLKernels.h
@@ -113,6 +113,7 @@
#include "arm_compute/core/CL/kernels/CLQuantizationLayerKernel.h"
#include "arm_compute/core/CL/kernels/CLROIAlignLayerKernel.h"
#include "arm_compute/core/CL/kernels/CLROIPoolingLayerKernel.h"
+#include "arm_compute/core/CL/kernels/CLRangeKernel.h"
#include "arm_compute/core/CL/kernels/CLReductionOperationKernel.h"
#include "arm_compute/core/CL/kernels/CLRemapKernel.h"
#include "arm_compute/core/CL/kernels/CLReorgLayerKernel.h"
diff --git a/arm_compute/core/CL/kernels/CLRangeKernel.h b/arm_compute/core/CL/kernels/CLRangeKernel.h
new file mode 100644
index 0000000..2da2117
--- /dev/null
+++ b/arm_compute/core/CL/kernels/CLRangeKernel.h
@@ -0,0 +1,83 @@
+/*
+ * Copyright (c) 2018 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __ARM_COMPUTE_CLRANGEKERNEL_H__
+#define __ARM_COMPUTE_CLRANGEKERNEL_H__
+
+#include "arm_compute/core/CL/ICLKernel.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+class ICLTensor;
+
+/** Kernel class for Range()
+ *
+ * range generates a 1-D tensor containing a sequence of numbers that begins at 'start' and extends by increments
+ * of 'step' up to but not including 'end'.
+ */
+class CLRangeKernel : public ICLKernel
+{
+public:
+ /** Default constructor */
+ CLRangeKernel();
+ /** Prevent instances of this class from being copied (As this class contains pointers) */
+ CLRangeKernel(const CLRangeKernel &) = delete;
+ /** Prevent instances of this class from being copied (As this class contains pointers) */
+ CLRangeKernel &operator=(const CLRangeKernel &) = delete;
+ /** Allow instances of this class to be moved */
+ CLRangeKernel(CLRangeKernel &&) = default;
+ /** Allow instances of this class to be moved */
+ CLRangeKernel &operator=(CLRangeKernel &&) = default;
+ /** Default destructor */
+ ~CLRangeKernel() = default;
+ /** Initialise the kernel's output tensor, start, end and step of the sequence.
+ *
+ * @param[out] output Output tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] start The starting value of the sequence.
+ * @param[in] end The ending (not including) value of the sequence.
+ * @param[in] step The gap between each pair of values in the sequence.
+ */
+ void configure(ICLTensor *output, float start, float end, float step);
+ /** Static function to check if given info will lead to a valid configuration of @ref CLRangeKernel
+ *
+ * @param[in] output Output tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] start The starting value of the sequence.
+ * @param[in] end The ending (not including) value of the sequence.
+ * @param[in] step The gap between each pair of values in the sequence.
+ *
+ * @return a status
+ */
+ static Status validate(const ITensorInfo *output, float start, float end, float step);
+
+ // Inherited methods overridden:
+ void run(const Window &window, cl::CommandQueue &queue) override;
+
+private:
+ float _start; /**< Start of sequence */
+ float _end; /**< End of sequence */
+ float _step; /**< Increment/step value */
+ ICLTensor *_output; /**< Destination tensor */
+};
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_CLRANGEKERNEL_H__ */
diff --git a/arm_compute/core/utils/misc/Utility.h b/arm_compute/core/utils/misc/Utility.h
index 37c8b66..8dd9afd 100644
--- a/arm_compute/core/utils/misc/Utility.h
+++ b/arm_compute/core/utils/misc/Utility.h
@@ -80,8 +80,10 @@
*
* @return Clamped value.
*/
-template <typename T>
-inline T clamp(const T &n, const T &lower, const T &upper)
+template <typename DataType, typename RangeType = DataType>
+inline DataType clamp(const DataType &n,
+ const DataType &lower = std::numeric_limits<RangeType>::lowest(),
+ const DataType &upper = std::numeric_limits<RangeType>::max())
{
return std::max(lower, std::min(n, upper));
}
diff --git a/arm_compute/runtime/CL/CLFunctions.h b/arm_compute/runtime/CL/CLFunctions.h
index d4827af..303a898 100644
--- a/arm_compute/runtime/CL/CLFunctions.h
+++ b/arm_compute/runtime/CL/CLFunctions.h
@@ -111,6 +111,7 @@
#include "arm_compute/runtime/CL/functions/CLRNNLayer.h"
#include "arm_compute/runtime/CL/functions/CLROIAlignLayer.h"
#include "arm_compute/runtime/CL/functions/CLROIPoolingLayer.h"
+#include "arm_compute/runtime/CL/functions/CLRange.h"
#include "arm_compute/runtime/CL/functions/CLReduceMean.h"
#include "arm_compute/runtime/CL/functions/CLReductionOperation.h"
#include "arm_compute/runtime/CL/functions/CLRemap.h"
diff --git a/arm_compute/runtime/CL/functions/CLRange.h b/arm_compute/runtime/CL/functions/CLRange.h
new file mode 100644
index 0000000..2614534
--- /dev/null
+++ b/arm_compute/runtime/CL/functions/CLRange.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2018 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __ARM_COMPUTE_CLRANGE_H__
+#define __ARM_COMPUTE_CLRANGE_H__
+
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/ICLSimpleFunction.h"
+
+namespace arm_compute
+{
+class ICLTensor;
+
+/** Basic function to run @ref CLRangeKernel
+ *
+ * @note The tensor data type for the output must be U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @note The function performs generates a sequence with the given start, end and step.
+ */
+class CLRange : public ICLSimpleFunction
+{
+public:
+ /** Initialise the kernel's start, end, step and output tensor.
+ *
+ * @param[out] output Output tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] start The starting value of the sequence.
+ * @param[in] end The ending (not including) value of the sequence.
+ * @param[in] step The gap between each pair of values in the sequence. Default is 1.
+ */
+ void configure(ICLTensor *output, float start, float end, float step = 1.f);
+ /** Static function to check if given info will lead to a valid configuration of @ref CLRange
+ *
+ * @param[in] output Output tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] start The starting value of the sequence.
+ * @param[in] end The ending (not including) value of the sequence.
+ * @param[in] step The gap between each pair of values in the sequence. Default is 1.
+ *
+ * @return a status
+ */
+ static Status validate(const ITensorInfo *output, float start, float end, float step = 1.f);
+};
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_CLRANGE_H__ */