COMPMID-472 : Implement Floor for CL and NEON.

Change-Id: I675a4545b1fe9ab665a07c834720bfe7ff589cee
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82527
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

6 files changed, 131 insertions, 0 deletions

diff --git a/arm_compute/core/CL/CLKernels.h b/arm_compute/core/CL/CLKernels.h
index cf36beffa3..0e64851457 100644
--- a/arm_compute/core/CL/CLKernels.h
+++ b/arm_compute/core/CL/CLKernels.h
@@ -50,6 +50,7 @@
 #include "arm_compute/core/CL/kernels/CLErodeKernel.h"
 #include "arm_compute/core/CL/kernels/CLFastCornersKernel.h"
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
+#include "arm_compute/core/CL/kernels/CLFloorKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h"
diff --git a/arm_compute/core/CL/kernels/CLFloorKernel.h b/arm_compute/core/CL/kernels/CLFloorKernel.h
new file mode 100644
index 0000000000..382c40ed39
--- /dev/null
+++ b/arm_compute/core/CL/kernels/CLFloorKernel.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2017 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_CLFLOORKERNEL_H__
+#define __ARM_COMPUTE_CLFLOORKERNEL_H__
+
+#include "arm_compute/core/CL/ICLKernel.h"
+
+namespace arm_compute
+{
+class ICLTensor;
+
+/** OpenCL kernel to perform a floor operation */
+class CLFloorKernel : public ICLKernel
+{
+public:
+    /** Default constructor */
+    CLFloorKernel();
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLFloorKernel(const CLFloorKernel &) = delete;
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLFloorKernel &operator=(const CLFloorKernel &) = delete;
+    /** Allow instances of this class to be moved */
+    CLFloorKernel(CLFloorKernel &&) = default;
+    /** Allow instances of this class to be moved */
+    CLFloorKernel &operator=(CLFloorKernel &&) = default;
+    /** Default destructor */
+    ~CLFloorKernel() = default;
+    /** Set the source, destination of the kernel
+     *
+     * @param[in]  input  Source tensor. Data type supported: F32.
+     * @param[out] output Destination tensor. Data type supported: F32.
+     */
+    void configure(const ICLTensor *input, ICLTensor *output);
+
+    // Inherited methods overridden:
+    void run(const Window &window, cl::CommandQueue &queue) override;
+
+private:
+    const ICLTensor *_input;
+    ICLTensor       *_output;
+};
+}
+#endif /*__ARM_COMPUTE_CLFLOORKERNEL_H__ */
diff --git a/arm_compute/core/NEON/NEKernels.h b/arm_compute/core/NEON/NEKernels.h
index ee63be3c46..a326f36b76 100644
--- a/arm_compute/core/NEON/NEKernels.h
+++ b/arm_compute/core/NEON/NEKernels.h
@@ -54,6 +54,7 @@
 #include "arm_compute/core/NEON/kernels/NEFillArrayKernel.h"
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
 #include "arm_compute/core/NEON/kernels/NEFillInnerBorderKernel.h"
+#include "arm_compute/core/NEON/kernels/NEFloorKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMLowpMatrixMultiplyKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMMatrixAccumulateBiasesKernel.h"
diff --git a/arm_compute/core/NEON/NEMath.h b/arm_compute/core/NEON/NEMath.h
index 39f0c3bf77..523649c65d 100644
--- a/arm_compute/core/NEON/NEMath.h
+++ b/arm_compute/core/NEON/NEMath.h
@@ -28,6 +28,14 @@
 
 namespace arm_compute
 {
+/** Calculate floor of a vector.
+ *
+ * @param[in] val Input vector value in F32 format.
+ *
+ * @return The calculated floor vector.
+ */
+float32x4_t vfloorq_f32(float32x4_t val);
+
 /** Calculate inverse square root.
  *
  * @param[in] x Input value.
diff --git a/arm_compute/core/NEON/NEMath.inl b/arm_compute/core/NEON/NEMath.inl
index 08f6749ac9..bdd747c4e9 100644
--- a/arm_compute/core/NEON/NEMath.inl
+++ b/arm_compute/core/NEON/NEMath.inl
@@ -54,6 +54,16 @@ const std::array<float32x4_t, 8> log_tab =
     }
 };
 
+inline float32x4_t vfloorq_f32(float32x4_t val)
+{
+    static const float32x4_t CONST_1 = vdupq_n_f32(1.f);
+
+    const int32x4_t   z = vcvtq_s32_f32(val);
+    const float32x4_t r = vcvtq_f32_s32(z);
+
+    return vbslq_f32(vcgtq_f32(r, val), vsubq_f32(r, CONST_1), r);
+}
+
 inline float32x4_t vinvsqrtq_f32(float32x4_t x)
 {
     float32x4_t sqrt_reciprocal = vrsqrteq_f32(x);
diff --git a/arm_compute/core/NEON/kernels/NEFloorKernel.h b/arm_compute/core/NEON/kernels/NEFloorKernel.h
new file mode 100644
index 0000000000..3b925b093b
--- /dev/null
+++ b/arm_compute/core/NEON/kernels/NEFloorKernel.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright (c) 2017 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_NEFLOORKERNEL_H__
+#define __ARM_COMPUTE_NEFLOORKERNEL_H__
+
+#include "arm_compute/core/NEON/INESimpleKernel.h"
+
+namespace arm_compute
+{
+class ITensor;
+
+/** NEON kernel to perform a floor operation */
+class NEFloorKernel : public INESimpleKernel
+{
+public:
+    /** Set the source, destination of the kernel
+     *
+     * @param[in]  input  Source tensor. Data type supported: F32.
+     * @param[out] output Destination tensor. Data type supported: F32.
+     */
+    void configure(const ITensor *input, ITensor *output);
+    // Inherited methods overridden:
+    void run(const Window &window) override;
+};
+}
+#endif /*__ARM_COMPUTE_NEFLOORKERNEL_H__ */