CPU Direct Execution: Implement Ceiling

Also, formatting fixes

src/ngraph/runtime/cpu/cpu_builder.cpp

@@ -94,6 +94,7 @@
 #include "ngraph/runtime/cpu/cpu_op_annotations.hpp"
 #include "ngraph/runtime/cpu/kernel/abs.hpp"
 #include "ngraph/runtime/cpu/kernel/add.hpp"
+#include "ngraph/runtime/cpu/kernel/ceil.hpp"
 #include "ngraph/runtime/cpu/kernel/multiply.hpp"
 #include "ngraph/runtime/cpu/kernel/relu.hpp"
 #include "ngraph/runtime/cpu/kernel/result.hpp"
@@ -230,6 +231,25 @@ namespace ngraph
                 functors.emplace_back(functor);
             }
 
+            template <>
+            void Builder::BUILDER_DECL(ngraph::op::Ceiling)
+            {
+                auto& functors = external_function->get_functors();
+                auto& tensor_data = external_function->get_tensor_data();
+                std::function<void(void*, void*, size_t)> kernel;
+
+                SELECT_KERNEL(kernel, out[0].get_element_type(), runtime::cpu::kernel::ceil);
+
+                auto element_count = out[0].get_size();
+                auto& arg0_tensor = tensor_data[args[0].get_name()];
+                auto& out0_tensor = tensor_data[out[0].get_name()];
+
+                auto functor = [&, kernel, element_count](CPURuntimeContext* ctx) {
+                    kernel(arg0_tensor, out0_tensor, element_count);
+                };
+                functors.emplace_back(functor);
+            }
+
             template <>
             void Builder::BUILDER_DECL(ngraph::op::Relu)
             {
@@ -307,8 +327,11 @@ namespace ngraph
 
                 const float beta = 0.0f;
 
-                auto mm_functor = [&, transpose_A, transpose_B, m, n, k, lda, ldb, beta, arg2_shape](CPURuntimeContext* ctx) {
-                                      cblas::cblas_sgemm(cblas::Layout::RowMajor,
+                auto mm_functor =
+                    [&, transpose_A, transpose_B, m, n, k, lda, ldb, beta, arg2_shape](
+                        CPURuntimeContext* ctx) {
+                        cblas::cblas_sgemm(
+                            cblas::Layout::RowMajor,
                             transpose_A ? cblas::Transpose::Transpose : cblas::Transpose::None,
                             transpose_B ? cblas::Transpose::Transpose : cblas::Transpose::None,
                             m,
@@ -321,8 +344,7 @@ namespace ngraph
                             max(1UL, ldb),
                             beta,
                             static_cast<float*>(out0_tensor),
-                                                         max(1UL, arg2_shape[1])
-                                          );
+                            max(1UL, arg2_shape[1]));
                     };
 
                 function<void(CPURuntimeContext*)> bias_functor = [](CPURuntimeContext* ctx) {};
@@ -351,8 +373,7 @@ namespace ngraph
                                                    max(1UL, arg2_shape[1]),
                                                    1.0f,
                                                    static_cast<float*>(out0_tensor),
-                                                                  max(1UL, arg2_shape[1])
-                                                   );
+                                                   max(1UL, arg2_shape[1]));
                             };
                         }
                         else
@@ -372,8 +393,7 @@ namespace ngraph
                                                    max(1UL, arg2_shape[1]),
                                                    1.0f,
                                                    static_cast<float*>(out0_tensor),
-                                                                  max(1UL, arg2_shape[1])
-                                                   );
+                                                   max(1UL, arg2_shape[1]));
                             };
                         }
                     }
@@ -401,8 +421,7 @@ namespace ngraph
                                                max(1UL, arg2_shape[1]),
                                                1.0f,
                                                static_cast<float*>(out0_tensor),
-                                                              max(1UL, arg2_shape[1])
-                                               );
+                                               max(1UL, arg2_shape[1]));
                         };
                     }
                 }
@@ -446,6 +465,7 @@ namespace ngraph
                 {TI(ngraph::op::Multiply), &runtime::cpu::Builder::build<ngraph::op::Multiply>},
                 {TI(ngraph::op::Parameter), &runtime::cpu::Builder::nop},
                 {TI(ngraph::op::Abs), &runtime::cpu::Builder::build<ngraph::op::Abs>},
+                {TI(ngraph::op::Ceiling), &runtime::cpu::Builder::build<ngraph::op::Ceiling>},
                 {TI(ngraph::op::Relu), &runtime::cpu::Builder::build<ngraph::op::Relu>},
                 {TI(ngraph::op::Result), &runtime::cpu::Builder::build<ngraph::op::Result>},
                 {TI(ngraph::op::MatmulBias), &runtime::cpu::Builder::build<ngraph::op::MatmulBias>},

src/ngraph/runtime/cpu/kernel/ceil.hpp

+/*******************************************************************************
+* Copyright 2018 Intel Corporation
+*
+* Licensed under the Apache License, Version 2.0 (the "License");
+* you may not use this file except in compliance with the License.
+* You may obtain a copy of the License at
+*
+*     http://www.apache.org/licenses/LICENSE-2.0
+*
+* Unless required by applicable law or agreed to in writing, software
+* distributed under the License is distributed on an "AS IS" BASIS,
+* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+* See the License for the specific language governing permissions and
+* limitations under the License.
+*******************************************************************************/
+
+#pragma once
+
+#define EIGEN_USE_THREADS
+#include <unsupported/Eigen/CXX11/Tensor>
+
+#include "ngraph/runtime/cpu/kernel/eigen_thread_pool.hpp"
+
+namespace ngraph
+{
+    namespace runtime
+    {
+        namespace cpu
+        {
+            namespace kernel
+            {
+                template <typename ElementType>
+                void ceil(void* input0, void* output, size_t count)
+                {
+                    Eigen::array<Eigen::Index, 1> out_dims, in_dims;
+
+                    out_dims[0] = in_dims[0] = count;
+
+                    Eigen::TensorMap<Eigen::Tensor<ElementType, 1, Eigen::RowMajor>> out(
+                        static_cast<ElementType*>(output), out_dims);
+                    Eigen::TensorMap<Eigen::Tensor<ElementType, 1, Eigen::RowMajor>> in0(
+                        static_cast<ElementType*>(input0), in_dims);
+
+                    out.device(eigen::global_thread_pool_device) = in0.ceil();
+                }
+            }
+        }
+    }
+}