Support super resolution & fix d2s opencl bugs

add neg/scalar_math

mace/core/operator.cc

@@ -82,12 +82,14 @@ extern void Register_FusedConv2D(OperatorRegistry *op_registry);
 extern void Register_GlobalAvgPooling(OperatorRegistry *op_registry);
 extern void Register_ImageToBuffer(OperatorRegistry *op_registry);
 extern void Register_MatMul(OperatorRegistry *op_registry);
+extern void Register_Neg(OperatorRegistry *op_registry);
 extern void Register_Pooling(OperatorRegistry *op_registry);
 extern void Register_Proposal(OperatorRegistry *op_registry);
 extern void Register_PSROIAlign(OperatorRegistry *op_registry);
 extern void Register_ReOrganize(OperatorRegistry *op_registry);
 extern void Register_Reshape(OperatorRegistry *op_registry);
 extern void Register_ResizeBilinear(OperatorRegistry *op_registry);
+extern void Register_ScalarMath(OperatorRegistry *op_registry);
 extern void Register_Slice(OperatorRegistry *op_registry);
 extern void Register_Softmax(OperatorRegistry *op_registry);
 extern void Register_SpaceToBatchND(OperatorRegistry *op_registry);
@@ -118,12 +120,14 @@ OperatorRegistry::OperatorRegistry() {
   ops::Register_GlobalAvgPooling(this);
   ops::Register_ImageToBuffer(this);
   ops::Register_MatMul(this);
+  ops::Register_Neg(this);
   ops::Register_Pooling(this);
   ops::Register_Proposal(this);
   ops::Register_PSROIAlign(this);
   ops::Register_ReOrganize(this);
   ops::Register_Reshape(this);
   ops::Register_ResizeBilinear(this);
+  ops::Register_ScalarMath(this);
   ops::Register_Slice(this);
   ops::Register_Softmax(this);
   ops::Register_SpaceToBatchND(this);

mace/kernels/eltwise.h

@@ -19,6 +19,7 @@ enum EltwiseType {
   SUM = 1,
   MAX = 2,
   MIN = 3,
+  SUB = 4,
 };
 
 struct EltwiseFunctorBase {
@@ -40,7 +41,7 @@ struct EltwiseFunctor : EltwiseFunctorBase {
                   StatsFuture *future) {
     Tensor::MappingGuard input0_guard(input0);
     Tensor::MappingGuard input1_guard(input1);
-    Tensor::MappingGuard output_guard(output);
+    Tensor::MappingGuard output_guard(output);    
 
     const T *input0_ptr = input0->data<T>();
     const T *input1_ptr = input1->data<T>();
@@ -51,35 +52,41 @@ struct EltwiseFunctor : EltwiseFunctorBase {
       case PROD:
 #pragma omp parallel for
         for (index_t i = 0; i < size; ++i) {
-          output_ptr[i] = input0_ptr[i] * input1_ptr[i];
+          output_ptr[i] =  input0_ptr[i] * input1_ptr[i];
         }
         break;
       case SUM:
-        if (coeff_.empty()) {
+        if (coeff_.empty()) {			
 #pragma omp parallel for
-          for (index_t i = 0; i < size; ++i) {
+	      for (index_t i = 0; i < size; ++i) {
             output_ptr[i] = input0_ptr[i] + input1_ptr[i];
           }
-        } else {
+        } else {		    
 #pragma omp parallel for
           for (index_t i = 0; i < size; ++i) {
             output_ptr[i] =
-                coeff_[0] * input0_ptr[i] + coeff_[1] * input1_ptr[i];
+              coeff_[0] * input0_ptr[i] + coeff_[1] * input1_ptr[i];
           }
         }
         break;
-      case MAX:
+      case MAX:        
 #pragma omp parallel for
         for (index_t i = 0; i < size; ++i) {
           output_ptr[i] = std::max<T>(input0_ptr[i], input1_ptr[i]);
         }
         break;
-      case MIN:
+      case MIN:        
 #pragma omp parallel for
         for (index_t i = 0; i < size; ++i) {
           output_ptr[i] = std::min<T>(input0_ptr[i], input1_ptr[i]);
         }
         break;
+      case SUB:
+#pragma omp parallel for
+        for (index_t i = 0; i < size; ++i) {
+          output_ptr[i] = input0_ptr[i] - input1_ptr[i];
+        }
+        break;
       default:
         LOG(FATAL) << "Eltwise op not support type " << type_;
     }

mace/kernels/negative.h

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#ifndef MACE_KERNELS_NEGATIVE_H_
+#define MACE_KERNELS_NEGATIVE_H_
+
+#include <vector>
+
+#include "mace/core/future.h"
+#include "mace/core/runtime/opencl/cl2_header.h"
+#include "mace/core/tensor.h"
+#include "mace/public/mace.h"
+
+namespace mace {
+namespace kernels {
+
+template <DeviceType D, typename T>
+struct NegFunctor {
+  void operator()(const Tensor *input,
+                  Tensor *output,
+                  StatsFuture *future) {
+    const index_t batch = input->dim(0);
+    const index_t height = input->dim(1);
+    const index_t width = input->dim(2);
+    const index_t channels = input->dim(3);
+
+    Tensor::MappingGuard input_mapper(input);
+    Tensor::MappingGuard output_mapper(output);
+
+    const T *input_ptr = input->data<T>();
+    T *output_ptr = output->mutable_data<T>();
+
+#pragma omp parallel for collapse(4)
+    for (index_t n = 0; n < batch; ++n) {
+      for (index_t h = 0; h < height; ++h) {
+        for (index_t w = 0; w < width; ++w) {
+          for (index_t c = 0; c < channels; ++c) {
+            index_t pos = (((n * height) + h) * width + w) * channels + c;
+            output_ptr[pos] = 0 - input_ptr[pos];
+          }
+        }
+      }
+    }
+  }
+};
+
+/*
+template <>
+void NegFunctor<DeviceType::NEON, float>::operator()(
+    const Tensor *input,
+    const Tensor *bias,
+    Tensor *output,
+    StatsFuture *future);
+*/
+
+template <typename T>
+struct NegFunctor<DeviceType::OPENCL, T> {
+  void operator()(const Tensor *input,
+                  Tensor *output,
+                  StatsFuture *future);
+  cl::Kernel kernel_;
+  std::vector<index_t> input_shape_;
+};
+
+}  // namespace kernels
+}  // namespace mace
+
+#endif  // MACE_KERNELS_NEGATIVE_H_

mace/kernels/opencl/cl/depth_to_space.cl

@@ -10,19 +10,16 @@ __kernel void depth_to_space(__read_only image2d_t input,
   const int output_width = get_global_size(1);
 
   const int out_pos = mad24(out_d, output_width, out_w);
-
   const int input_width = output_width / block_size;
 
   const int in_h = out_h / block_size;
   const int offset_h = out_h % block_size;
   const int in_w = out_w / block_size;
   const int offset_w = out_w % block_size;
-
   const int offset_d = (offset_h * block_size + offset_w) * output_depth;
   const int in_d = out_d + offset_d;
 
   const int in_pos = mad24(in_d, input_width, in_w);
-
   DATA_TYPE4 in_data = READ_IMAGET(input, SAMPLER, (int2)(in_pos, in_h));
   WRITE_IMAGET(output, (int2)(out_pos, out_h), in_data);
 }

mace/kernels/opencl/cl/eltwise.cl

@@ -27,8 +27,9 @@ __kernel void eltwise(__read_only image2d_t input0, /* [c%4 * w * c/4, h * b] */
   out = fmax(in0, in1);
 #elif ELTWISE_TYPE == 3
   out = fmin(in0, in1);
+#elif ELTWISE_TYPE == 4
+  out = in0 - in1;
 #endif
 
   WRITE_IMAGET(output, (int2)(w, hb), out);
 }
-

mace/kernels/opencl/cl/neg.cl

+#include <common.h>
+// Supported data types: half/float
+__kernel void neg(__read_only image2d_t input,
+                       __write_only image2d_t output) {
+  const int ch_blk = get_global_id(0);
+  const int w = get_global_id(1);
+  const int hb = get_global_id(2);
+  const int width = get_global_size(1);
+
+  const int pos = mad24(ch_blk, width, w);
+  DATA_TYPE4 in = READ_IMAGET(input, SAMPLER, (int2)(pos, hb));
+  DATA_TYPE4 out = 0 - in;
+  WRITE_IMAGET(output, (int2)(pos, hb), out);
+}

mace/kernels/opencl/cl/scalar_math.cl

+#include <common.h>
+
+__kernel void scalar_math(__read_only image2d_t input, /* [c%4 * w * c/4, h * b] */
+                      __private const float scalar,
+                      __write_only image2d_t output) {
+  const int w = get_global_id(0);
+  const int hb = get_global_id(1);
+
+  DATA_TYPE4 in0 = READ_IMAGET(input, SAMPLER, (int2)(w, hb));
+  DATA_TYPE4 in1;
+  in1.x = scalar;
+  in1.y = scalar;
+  in1.z = scalar;
+  in1.w = scalar;  
+  DATA_TYPE4 out;
+#if SCALAR_MATH_TYPE == 1
+  out = in0 + in1;
+#elif SCALAR_MATH_TYPE == 4
+  out = in0 - in1;
+#elif SCALAR_MATH_TYPE == 0
+  out = in0 * in1;
+#elif SCALAR_MATH_TYPE == 5
+  out = in0 / in1;
+#endif
+
+  WRITE_IMAGET(output, (int2)(w, hb), out);
+}

mace/kernels/opencl/depth_to_space_opencl.cc

@@ -22,6 +22,7 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
 
   int depth_blocks = 1;
   const char *kernel_name = nullptr;
+  index_t kernel_width = input_width;
 
   index_t output_height, output_width, output_depth;
   if (d2s_) {
@@ -30,12 +31,14 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
     output_depth = input_depth / (block_size_ * block_size_);
     depth_blocks = RoundUpDiv4(output_depth);
     kernel_name = "depth_to_space";
+    kernel_width = output_width;
   } else {
     output_height = input_height / block_size_;
     output_width = input_width / block_size_;
     output_depth = input_depth * block_size_ * block_size_;
     depth_blocks = RoundUpDiv4(input_depth);
     kernel_name = "space_to_depth";
+    kernel_width = input_width;
   }
 
   std::vector<index_t> output_shape = {batch, output_height, output_width,
@@ -53,16 +56,17 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
     kernel_name_ss << "-D" << kernel_name << "=" << obfuscated_kernel_name;
     built_options.emplace(kernel_name_ss.str());
     auto dt = DataTypeToEnum<T>::value;
-    built_options.emplace("-DDATA_TYPE=" + DtToUpstreamCLDt(dt));
-    built_options.emplace("-DCMD_DATA_TYPE=" + DtToUpstreamCLCMDDt(dt));
+    built_options.emplace("-DDATA_TYPE=" + DtToCLDt(dt));
+    built_options.emplace("-DCMD_DATA_TYPE=" + DtToCLCMDDt(dt));
     kernel_ =
-        runtime->BuildKernel("depth_to_space", kernel_name, built_options);
+        runtime->BuildKernel("depth_to_space",
+                             obfuscated_kernel_name, built_options);
   }
   if (!IsVecEqual(input_shape_, input->shape())) {
     uint32_t idx = 0;
     kernel_.setArg(idx++, *(input->opencl_image()));
-    kernel_.setArg(idx++, block_size_);
-    kernel_.setArg(idx++, depth_blocks);
+    kernel_.setArg(idx++, static_cast<int32_t>(block_size_));
+    kernel_.setArg(idx++, static_cast<int32_t>(depth_blocks));
     kernel_.setArg(idx++, *(output->opencl_image()));
     input_shape_ = input->shape();
   }
@@ -74,8 +78,7 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
     const std::vector<uint32_t> lws = {8, 16, 8, 1};
     std::stringstream ss;
     ss << "depth_to_space_opencl_kernel_" << output->dim(0) << "_"
-       << output->dim(1) << "_" << output->dim(2) << "_" << output->dim(3);
-
+       << output->dim(1) << "_" << output->dim(2) << "_" << depth_blocks;
     TuningOrRun3DKernel(kernel_, ss.str(), gws, lws, future);
   } else {
     const uint32_t gws[3] = {static_cast<uint32_t>(depth_blocks),
@@ -83,8 +86,8 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
                              static_cast<uint32_t>(input_height * batch)};
     const std::vector<uint32_t> lws = {8, 16, 8, 1};
     std::stringstream ss;
-    ss << "space_to_depth_opencl_kernel_" << input->dim(0) << "_"
-       << input->dim(1) << "_" << input->dim(2) << "_" << input->dim(3);
+    ss << "depth_to_space_opencl_kernel_" << input->dim(0) << "_"
+       << input->dim(1) << "_" << input->dim(2) << "_" << depth_blocks;
     TuningOrRun3DKernel(kernel_, ss.str(), gws, lws, future);
   }
 }

mace/kernels/opencl/eltwise_opencl.cc

@@ -22,7 +22,7 @@ void EltwiseFunctor<DeviceType::OPENCL, T>::operator()(const Tensor *input0,
 
   const index_t channel_blocks = RoundUpDiv4(channels);
   const index_t width_pixels = channel_blocks * width;
-  const index_t batch_height_pixels = batch * height;
+  const index_t batch_height_pixels = batch * height;  
 
   if (kernel_.get() == nullptr) {
     auto runtime = OpenCLRuntime::Global();

mace/kernels/opencl/negative_opencl.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/kernels/negative.h"
+#include "mace/core/runtime/opencl/cl2_header.h"
+#include "mace/core/runtime/opencl/opencl_runtime.h"
+#include "mace/kernels/opencl/helper.h"
+#include "mace/utils/utils.h"
+
+namespace mace {
+namespace kernels {
+
+template <typename T>
+void NegFunctor<DeviceType::OPENCL, T>::operator()(const Tensor *input,
+                                                       Tensor *output,
+                                                       StatsFuture *future) {
+  const index_t batch = input->dim(0);
+  const index_t height = input->dim(1);
+  const index_t width = input->dim(2);
+  const index_t channels = input->dim(3);
+
+  const index_t channel_blocks = RoundUpDiv4(channels);
+
+  auto runtime = OpenCLRuntime::Global();
+  if (kernel_.get() == nullptr) {
+    std::set<std::string> built_options;
+    auto dt = DataTypeToEnum<T>::value;
+    std::string kernel_name = MACE_OBFUSCATE_SYMBOL("neg");
+    built_options.emplace("-Dneg=" + kernel_name);
+    built_options.emplace("-DDATA_TYPE=" + DtToUpstreamCLDt(dt));
+    built_options.emplace("-DCMD_DATA_TYPE=" + DtToUpstreamCLCMDDt(dt));
+    kernel_ = runtime->BuildKernel("neg", kernel_name, built_options);
+  }
+  if (!IsVecEqual(input_shape_, input->shape())) {
+    uint32_t idx = 0;
+    kernel_.setArg(idx++, *(input->opencl_image()));
+    kernel_.setArg(idx++, *(output->opencl_image()));
+    input_shape_ = input->shape();
+  }
+
+  const uint32_t gws[3] = {static_cast<uint32_t>(channel_blocks),
+                           static_cast<uint32_t>(width),
+                           static_cast<uint32_t>(height * batch)};
+  const std::vector<uint32_t> lws = {8, 16, 8};
+
+  cl::Event event;
+  cl_int error = runtime->command_queue().enqueueNDRangeKernel(
+      kernel_, cl::NullRange, cl::NDRange(gws[0], gws[1], gws[2]),
+      cl::NDRange(lws[0], lws[1], lws[2]), nullptr, &event);
+  MACE_CHECK(error == CL_SUCCESS);
+  if (future != nullptr) {
+    future->wait_fn = [runtime, event](CallStats *stats) {
+      event.wait();
+      if (stats != nullptr) {
+        runtime->GetCallStats(event, stats);
+      }
+    };
+  }
+}
+
+template struct NegFunctor<DeviceType::OPENCL, float>;
+template struct NegFunctor<DeviceType::OPENCL, half>;
+}  // namespace kernels
+}  // namespace mace

mace/kernels/opencl/scalar_math_opencl.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/kernels/scalar_math.h"
+#include "mace/core/runtime/opencl/opencl_runtime.h"
+#include "mace/kernels/opencl/helper.h"
+#include "mace/utils/tuner.h"
+
+namespace mace {
+namespace kernels {
+
+template <typename T>
+void ScalarMathFunctor<DeviceType::OPENCL, T>::operator()(const Tensor *input,
+                                                       Tensor *output,
+                                                       StatsFuture *future) {
+  const index_t batch = input->dim(0);
+  const index_t height = input->dim(1);
+  const index_t width = input->dim(2);
+  const index_t channels = input->dim(3);
+
+  const index_t channel_blocks = RoundUpDiv4(channels);
+  const index_t width_pixels = channel_blocks * width;
+  const index_t batch_height_pixels = batch * height;
+
+  if (kernel_.get() == nullptr) {
+    auto runtime = OpenCLRuntime::Global();
+    std::set<std::string> built_options;
+    auto dt = DataTypeToEnum<T>::value;
+    std::string kernel_name = MACE_OBFUSCATE_SYMBOL("scalar_math");
+    built_options.emplace("-Dscalar_math=" + kernel_name);
+    built_options.emplace("-DDATA_TYPE=" + DtToUpstreamCLDt(dt));
+    built_options.emplace("-DCMD_DATA_TYPE=" + DtToUpstreamCLCMDDt(dt));
+    built_options.emplace(MakeString("-DSCALAR_MATH_TYPE=", type_));
+    kernel_ = runtime->BuildKernel("scalar_math", kernel_name, built_options);
+  }
+  if (!IsVecEqual(input_shape_, input->shape())) {
+    uint32_t idx = 0;
+    kernel_.setArg(idx++, *(input->opencl_image()));
+    kernel_.setArg(idx++, static_cast<float>(coeff_));
+    kernel_.setArg(idx++, *(output->opencl_image()));
+    input_shape_ = input->shape();
+  }
+
+  const uint32_t gws[2] = {static_cast<uint32_t>(width_pixels),
+                           static_cast<uint32_t>(batch_height_pixels)};
+  const std::vector<uint32_t> lws = {64, 16, 1};
+  std::stringstream ss;
+  ss << "eltwise_opencl_kernel_" << output->dim(0) << "_" << output->dim(1)
+     << "_" << output->dim(2) << "_" << output->dim(3);
+  TuningOrRun2DKernel(kernel_, ss.str(), gws, lws, future);
+}
+
+template struct ScalarMathFunctor<DeviceType::OPENCL, float>;
+template struct ScalarMathFunctor<DeviceType::OPENCL, half>;
+}  // namespace kernels
+}  // namespace mace

mace/kernels/opencl/space_to_batch_opencl.cc

@@ -42,7 +42,7 @@ void SpaceToBatchFunctor<DeviceType::OPENCL, T>::operator()(
     built_options.emplace("-DCMD_DATA_TYPE=" +
                           DtToCLCMDDt(DataTypeToEnum<T>::value));
     kernel_ =
-        runtime->BuildKernel("space_to_batch", kernel_name, built_options);
+        runtime->BuildKernel("space_to_batch", obfuscated_kernel_name, built_options);
   }
   if (!IsVecEqual(space_shape_, space_tensor->shape())) {
     uint32_t idx = 0;

mace/kernels/scalar_math.h

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+#ifndef MACE_KERNELS_SCALAR_MATH_H_
+#define MACE_KERNELS_SCALAR_MATH_H_
+
+#include <algorithm>
+#include <vector>
+
+#include "mace/core/future.h"
+#include "mace/core/runtime/opencl/cl2_header.h"
+#include "mace/core/tensor.h"
+
+namespace mace {
+namespace kernels {
+
+enum ScalarMathType {
+  MUL = 0,
+  ADD = 1,
+  MAX = 2,
+  MIN = 3,
+  SUB = 4,
+  DIV = 5,
+};
+
+struct ScalarMathFunctorBase {
+  ScalarMathFunctorBase(const ScalarMathType type, const float coeff)
+      : type_(type), coeff_(coeff) {}
+
+  ScalarMathType type_;
+  float coeff_;
+};
+
+template <DeviceType D, typename T>
+struct ScalarMathFunctor : ScalarMathFunctorBase {
+  ScalarMathFunctor(const ScalarMathType type, const float coeff)
+      : ScalarMathFunctorBase(type, coeff) {}
+
+  void operator()(const Tensor *input,
+                  Tensor *output,
+                  StatsFuture *future) {
+    Tensor::MappingGuard input_guard(input);
+    Tensor::MappingGuard output_guard(output);
+
+    const T *input_ptr = input->data<T>();
+    T *output_ptr = output->mutable_data<T>();
+    const index_t size = input->size();
+
+    switch (type_) {
+      case MUL:
+#pragma omp parallel for
+        for (index_t i = 0; i < size; ++i) {
+          output_ptr[i] =  coeff_ * input_ptr[i];
+        }
+        break;
+      case ADD:
+#pragma omp parallel for
+        for (index_t i = 0; i < size; ++i) {
+          output_ptr[i] =  coeff_ + input_ptr[i];
+        }
+        break;
+      case SUB:
+#pragma omp parallel for
+        for (index_t i = 0; i < size; ++i) {
+          output_ptr[i] =  input_ptr[i] - coeff_;
+        }
+        break;
+      case DIV:
+#pragma omp parallel for
+        for (index_t i = 0; i < size; ++i) {
+          output_ptr[i] =  input_ptr[i] / coeff_;
+        }
+        break;
+      default:
+        LOG(FATAL) << "ScalarMath op not support type " << type_;
+    }
+  }
+};
+
+template <typename T>
+struct ScalarMathFunctor<DeviceType::OPENCL, T> : ScalarMathFunctorBase {
+  ScalarMathFunctor(const ScalarMathType type, const float coeff)
+      : ScalarMathFunctorBase(type, coeff) {}
+
+  void operator()(const Tensor *input,
+                  Tensor *output,
+                  StatsFuture *future);
+
+  cl::Kernel kernel_;
+  std::vector<index_t> input_shape_;
+};
+
+}  // namespace kernels
+}  // namespace mace
+
+#endif  // MACE_KERNELS_SCALAR_MATH_H_

mace/ops/conv_2d.h

@@ -31,7 +31,7 @@ class Conv2dOp : public ConvPool2dOpBase<D, T> {
     const Tensor *filter = this->Input(FILTER);
     const Tensor *bias = this->InputSize() >= 3 ? this->Input(BIAS) : nullptr;
     Tensor *output = this->Output(OUTPUT);
-
+    
     functor_(input, filter, bias, output, future);
 
     return true;

mace/ops/depth_to_space.h

@@ -19,18 +19,16 @@ class DepthToSpaceOp : public Operator<D, T> {
  public:
   DepthToSpaceOp(const OperatorDef &op_def, Workspace *ws)
       : Operator<D, T>(op_def, ws),
-        functor_(OperatorBase::GetSingleArgument<int>("block_size", 1), true) {}
+        block_size_(OperatorBase::GetSingleArgument<int>("block_size", 1)),
+        functor_(this->block_size_, true) {}
 
   bool Run(StatsFuture *future) override {
     const Tensor *input = this->Input(INPUT);
     Tensor *output = this->Output(OUTPUT);
     MACE_CHECK(input->dim_size() == 4, "input dim should be 4");
 
-    const int block_size =
-        OperatorBase::GetSingleArgument<int>("block_size", 1);
-
     int input_depth = input->dim(3);
-    MACE_CHECK(input_depth % (block_size * block_size) == 0,
+    MACE_CHECK(input_depth % (block_size_ * block_size_) == 0,
                "input depth should be dividable by block_size * block_size",
                input->dim(3));
     MACE_CHECK((input_depth % 4) == 0,
@@ -40,6 +38,7 @@ class DepthToSpaceOp : public Operator<D, T> {
   }
 
  protected:
+  const int block_size_;
   OP_INPUT_TAGS(INPUT);
   OP_OUTPUT_TAGS(OUTPUT);
 

mace/ops/depth_to_space_test.cc

 //
 // Copyright (c) 2017 XiaoMi All rights reserved.
 //
+#include <fstream>
 
+#include <vector>
 #include "mace/core/operator.h"
 #include "mace/ops/ops_test_util.h"
 
@@ -48,6 +50,7 @@ void RunDepthToSpace(const bool d2s,
   ExpectTensorNear<float>(*expected, *net.GetOutput("Output"), 0.001);
 }
 
+
 class SpaceToDepthOpTest : public OpsTestBase {};
 
 TEST_F(SpaceToDepthOpTest, Input2x4x4_B2_CPU) {
@@ -70,6 +73,8 @@ TEST_F(SpaceToDepthOpTest, Input2x4x4_B2_OPENCL) {
       16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31});
 }
 
+
+
 TEST_F(SpaceToDepthOpTest, Input2x2x4_B2_CPU) {
   RunDepthToSpace<DeviceType::CPU>(false, {1, 2, 2, 4},
       {1, 2, 3, 4, 5, 6, 7, 8,
@@ -132,46 +137,83 @@ TEST_F(DepthToSpaceOpTest, Input1x1x16_B2_OPENCL) {
        9, 10, 11, 12, 13, 14, 15, 16});
 }
 
-/*
-TEST_F(DepthToSpaceOpTest, Input2x2x3_B2_CPU) {
 
-  RunDepthToSpace<DeviceType::CPU>({1, 2, 2, 3},
-                                   {1, 2, 3, 4, 5, 6,
-									7, 8, 9, 10, 11, 12},
-                                   2,
-                                   {1, 1, 1, 12},
-                                   {1,  2,  3,  4,  5,  6,  7,  8,  
-									9,  10, 11, 12});
+TEST_F(DepthToSpaceOpTest, InputLarger_B2_OPENCL) {
+  const std::vector<float > in = std::vector<float >(192 * 192 *128, 1.0);
+
+  RunDepthToSpace<DeviceType::OPENCL>(true, {1, 192, 192, 128},
+                                      in,
+                                      2,
+                                      {1, 384, 384, 32},
+                                      in);
+}
+
+
+template <DeviceType D, typename T>
+void RandomTest(const bool d2s, const int block_size,
+                const std::vector<index_t> &shape) {
+  testing::internal::LogToStderr();
+  srand(time(NULL));
+
+  // Construct graph
+  OpsTestNet net;
+
+  const char *ops_name = (d2s) ? "DepthToSpace" : "SpaceToDepth";
+  const char *ops_test_name = (d2s) ? "DepthToSpaceTest" : "SpaceToDepthTest";
+
+  // Add input data
+  net.AddRandomInput<D, float>("Input1", shape);
+
+  OpDefBuilder(ops_name, ops_test_name)
+      .Input("Input1")
+      .AddIntArg("block_size", block_size)
+      .Output("Output")
+      .Finalize(net.NewOperatorDef());
+
+  // Run
+  net.RunOp();
+
+  BufferToImage<D, T>(&net, "Input1", "InputImg1",
+                      kernels::BufferType::IN_OUT_CHANNEL);
+
+  OpDefBuilder(ops_name, ops_test_name)
+      .Input("InputImg1")
+      .AddIntArg("block_size", block_size)
+      .AddIntArg("T", static_cast<int>(DataTypeToEnum<T>::value))
+      .Output("OutputImg")
+      .Finalize(net.NewOperatorDef());
+
+  // Run
+  net.RunOp(D);
+
+  ImageToBuffer<D, float>(&net, "OutputImg", "OPENCLOutput",
+                          kernels::BufferType::IN_OUT_CHANNEL);
+
+  if (DataTypeToEnum<T>::value == DT_FLOAT) {
+    ExpectTensorNear<float>(*net.GetTensor("Output"),
+                            *net.GetOutput("OPENCLOutput"), 1e-3);
+  } else {
+    ExpectTensorNear<float>(*net.GetTensor("Output"),
+                            *net.GetOutput("OPENCLOutput"), 1e-1);
+  }
 }
 
-TEST_F(DepthToSpaceOpTest, Input2x2x3_B2_OPENCL) {
-  RunDepthToSpace<DeviceType::OPENCL>({1, 2, 2, 6},
-                                   {1, 2, 3, 4, 5, 6,
-									7, 8, 9, 10, 11, 12
-									},
-                                   2,
-                                   {1, 1, 1, 12},
-                                   {1,  2,  3,  4,  5,  6,  7,  8,  
-									9,  10, 11, 12});
+TEST_F(DepthToSpaceOpTest, OPENCLRandomFloat) {
+  RandomTest<DeviceType::OPENCL, float>(true, 2, {1, 192, 192, 128});
 }
 
-TEST_F(DepthToSpaceOpTest, Input2x2x2_B2_CPU) {
+TEST_F(DepthToSpaceOpTest, OPENCLRandomHalf) {
+RandomTest<DeviceType::OPENCL, half>(true, 2, {1, 192, 192, 128});
+}
 
-  RunDepthToSpace<DeviceType::CPU>({1, 2, 2, 2},
-                                   {1, 10,  2,  20,  3,  30,  4, 40},
-                                   2,
-                                   {1, 1, 1, 8},
-                                   {1, 10,  2,  20,  3,  30,  4, 40});
+TEST_F(SpaceToDepthOpTest, OPENCLRandomFloat) {
+RandomTest<DeviceType::OPENCL, float>(false, 2, {1, 384, 384, 32});
 }
 
-TEST_F(DepthToSpaceOpTest, Input2x2x2_B2_OPENCL) {
+TEST_F(SpaceToDepthOpTest, OPENCLRandomHalf) {
+RandomTest<DeviceType::OPENCL, half>(false, 2, {1, 384, 384, 32});
+}
 
-  RunDepthToSpace<DeviceType::OPENCL>({1, 2, 2, 2},
-                                      {1, 10,  2,  20,  3,  30,  4, 40},
-                                      2,
-                                      {1, 1, 1, 8},
-                                      {1, 10,  2,  20,  3,  30,  4, 40});
-}*/
 }  // namespace test
 }  // namespace ops
 }  // namespace mace

mace/ops/neg.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/ops/neg.h"
+
+namespace mace {
+namespace ops {
+
+void Register_Neg(OperatorRegistry *op_registry) {
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("Neg")
+                                     .Device(DeviceType::CPU)
+                                     .TypeConstraint<float>("T")
+                                     .Build(),
+                    NegOp<DeviceType::CPU, float>);
+
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("Neg")
+                                     .Device(DeviceType::OPENCL)
+                                     .TypeConstraint<float>("T")
+                                     .Build(),
+                    NegOp<DeviceType::OPENCL, float>);
+
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("Neg")
+                                     .Device(DeviceType::OPENCL)
+                                     .TypeConstraint<half>("T")
+                                     .Build(),
+                    NegOp<DeviceType::OPENCL, half>);
+}
+
+}  // namespace ops
+}  // namespace mace

mace/ops/neg.h

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#ifndef MACE_OPS_NEG_H_
+#define MACE_OPS_NEG_H_
+
+#include <string>
+
+#include "mace/core/operator.h"
+#include "mace/kernels/negative.h"
+
+namespace mace {
+namespace ops {
+
+template <DeviceType D, class T>
+class NegOp : public Operator<D, T> {
+ public:
+  NegOp(const OperatorDef &operator_def, Workspace *ws)
+      : Operator<D, T>(operator_def, ws),
+        functor_() {}
+
+  bool Run(StatsFuture *future) override {
+    const Tensor *input_tensor = this->Input(0);    
+    Tensor *output_tensor = this->outputs_[0];
+    output_tensor->ResizeLike(input_tensor);
+
+    functor_(input_tensor, output_tensor, future);
+    return true;
+  }
+
+ private:
+  kernels::NegFunctor<D, T> functor_;
+};
+
+}  // namespace ops
+}  // namespace mace
+
+#endif  // MACE_OPS_NEGATIVE_H_

mace/ops/neg_benchmark.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/core/operator.h"
+#include "mace/core/runtime/opencl/opencl_runtime.h"
+#include "mace/core/testing/test_benchmark.h"
+#include "mace/ops/ops_test_util.h"
+
+namespace mace {
+namespace ops {
+namespace test {
+
+template <DeviceType D, typename T>
+static void Neg(int iters, int batch, int channels, int height, int width) {
+  mace::testing::StopTiming();
+
+  OpsTestNet net;
+
+  // Add input data
+  net.AddRandomInput<D, T>("Input", {batch, height, width, channels});
+
+  if (D == DeviceType::OPENCL) {
+    BufferToImage<D, T>(&net, "Input", "InputImage",
+                        kernels::BufferType::IN_OUT_CHANNEL);
+   
+    OpDefBuilder("Neg", "NegBM")
+        .Input("InputImage")
+        .Output("Output")
+        .Finalize(net.NewOperatorDef());
+  } else {
+    OpDefBuilder("Neg", "NegBM")
+        .Input("Input")
+        .Output("Output")
+        .Finalize(net.NewOperatorDef());
+  }
+
+  // Warm-up
+  for (int i = 0; i < 5; ++i) {
+    net.RunOp(D);
+  }
+  net.Sync();
+
+  mace::testing::StartTiming();
+  while (iters--) {
+    net.RunOp(D);
+  }
+  net.Sync();
+}
+
+#define BM_NEG_MACRO(N, C, H, W, TYPE, DEVICE)                  \
+  static void BM_NEG_##N##_##C##_##H##_##W##_##TYPE##_##DEVICE( \
+      int iters) {                                                   \
+    const int64_t tot = static_cast<int64_t>(iters) * N * C * H * W; \
+    mace::testing::MaccProcessed(tot);                               \
+    mace::testing::BytesProcessed(tot *(sizeof(TYPE)));              \
+    Neg<DEVICE, TYPE>(iters, N, C, H, W);                        \
+  }                                                                  \
+  BENCHMARK(BM_NEG_##N##_##C##_##H##_##W##_##TYPE##_##DEVICE)
+
+#define BM_NEG(N, C, H, W)                 \
+  BM_NEG_MACRO(N, C, H, W, float, CPU);    \
+  BM_NEG_MACRO(N, C, H, W, float, OPENCL); \
+  BM_NEG_MACRO(N, C, H, W, half, OPENCL);
+
+BM_NEG(1, 1, 512, 512);
+BM_NEG(1, 3, 128, 128);
+BM_NEG(1, 3, 512, 512);
+BM_NEG(1, 32, 112, 112);
+BM_NEG(1, 64, 256, 256);
+BM_NEG(1, 64, 512, 512);
+BM_NEG(1, 128, 56, 56);
+BM_NEG(1, 128, 256, 256);
+BM_NEG(1, 256, 14, 14);
+BM_NEG(1, 512, 14, 14);
+BM_NEG(1, 1024, 7, 7);
+BM_NEG(32, 1, 256, 256);
+BM_NEG(32, 3, 256, 256);
+
+}  // namespace test
+}  // namespace ops
+}  // namespace mace

mace/ops/neg_test.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/core/operator.h"
+#include "mace/ops/ops_test_util.h"
+
+namespace mace {
+namespace ops {
+namespace test {
+
+class NegOpTest : public OpsTestBase {};
+
+template <DeviceType D>
+void NegSimple() {
+  OpsTestNet net;
+
+  // Add input data
+  net.AddInputFromArray<D, float>("Input", {1, 6, 2, 1},
+                                  {5, 5, 7, 7, 9, 9, 11, 11, 13, 13, 15, 15});
+
+  if (D == DeviceType::OPENCL) {
+    BufferToImage<D, float>(&net, "Input", "InputImage",
+                            kernels::BufferType::IN_OUT_CHANNEL);
+    
+    OpDefBuilder("Neg", "NegTest")
+        .Input("InputImage")
+        .Output("OutputImage")
+        .Finalize(net.NewOperatorDef());
+    // Run
+    net.RunOp(D);
+
+    // Transfer output
+    ImageToBuffer<D, float>(&net, "OutputImage", "Output",
+                            kernels::BufferType::IN_OUT_CHANNEL);
+  } else {
+    OpDefBuilder("Neg", "NegTest")
+        .Input("Input")
+        .Output("Output")
+        .Finalize(net.NewOperatorDef());
+    // Run
+    net.RunOp(D);
+  }
+
+  // Check
+  auto expected = CreateTensor<float>(
+      {1, 6, 2, 1},
+      {-5, -5, -7, -7, -9, -9, -11, -11, -13, -13, -15, -15});
+
+  ExpectTensorNear<float>(*expected, *net.GetOutput("Output"), 1e-8);
+}
+
+TEST_F(NegOpTest, NegSimpleCPU) { NegSimple<DeviceType::CPU>(); }
+
+TEST_F(NegOpTest, NegSimpleOPENCL) {
+  NegSimple<DeviceType::OPENCL>();
+}
+
+}  // namespace test
+}  // namespace ops
+}  // namespace mace

mace/ops/scalar_math.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/ops/scalar_math.h"
+
+namespace mace {
+namespace ops {
+
+void Register_ScalarMath(OperatorRegistry *op_registry) {
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("ScalarMath")
+                                     .Device(DeviceType::CPU)
+                                     .TypeConstraint<float>("T")
+                                     .Build(),
+                    ScalarMathOp<DeviceType::CPU, float>);
+
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("ScalarMath")
+                                     .Device(DeviceType::OPENCL)
+                                     .TypeConstraint<float>("T")
+                                     .Build(),
+                    ScalarMathOp<DeviceType::OPENCL, float>);
+
+  REGISTER_OPERATOR(op_registry, OpKeyBuilder("ScalarMath")
+                                     .Device(DeviceType::OPENCL)
+                                     .TypeConstraint<half>("T")
+                                     .Build(),
+                    ScalarMathOp<DeviceType::OPENCL, half>);
+}
+
+}  // namespace ops
+}  // namespace mace

mace/ops/scalar_math.h

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#ifndef MACE_OPS_SCALAR_MATH_H_
+#define MACE_OPS_SCALAR_MATH_H_
+
+#include <string>
+
+#include "mace/core/operator.h"
+#include "mace/kernels/scalar_math.h"
+
+namespace mace {
+namespace ops {
+
+template <DeviceType D, class T>
+class ScalarMathOp : public Operator<D, T> {
+ public:
+  ScalarMathOp(const OperatorDef &operator_def, Workspace *ws)
+      : Operator<D, T>(operator_def, ws),
+        x_(OperatorBase::GetSingleArgument<float>("x", 1.0)),
+        functor_(static_cast<kernels::ScalarMathType>(
+                     OperatorBase::GetSingleArgument<int>(
+                         "type", static_cast<int>(
+                             kernels::ScalarMathType::ADD))),
+                 this->x_) {}
+
+  bool Run(StatsFuture *future) override {
+    const Tensor *input_tensor = this->Input(INPUT);
+    Tensor *output_tensor = this->Output(OUTPUT);
+    output_tensor->ResizeLike(input_tensor);
+
+    functor_(input_tensor, output_tensor, future);
+    return true;
+  }
+
+ protected:
+  const float x_;
+  OP_INPUT_TAGS(INPUT);
+  OP_OUTPUT_TAGS(OUTPUT);
+
+ private:
+  kernels::ScalarMathFunctor<D, T> functor_;
+};
+
+}  // namespace ops
+}  // namespace mace
+
+#endif  // MACE_OPS_SCALAR_MATH_H_

mace/ops/scalar_math_benchmark.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/core/operator.h"
+#include "mace/core/runtime/opencl/opencl_runtime.h"
+#include "mace/core/testing/test_benchmark.h"
+#include "mace/ops/ops_test_util.h"
+
+namespace mace {
+namespace ops {
+namespace test {
+
+template <DeviceType D, typename T>
+static void ScalarMath(int iters, int batch, int channels,
+                       int height, int width, float x, int type) {
+  mace::testing::StopTiming();
+
+  OpsTestNet net;
+
+  // Add input data
+  net.AddRandomInput<D, T>("Input", {batch, height, width, channels});
+
+  if (D == DeviceType::OPENCL) {
+    BufferToImage<D, T>(&net, "Input", "InputImage",
+                        kernels::BufferType::IN_OUT_CHANNEL);
+    OpDefBuilder("ScalarMath", "ScalarMathBM")
+        .Input("InputImage")
+        .Output("Output")
+        .AddIntArg("type", type)
+        .AddFloatArg("x", x)
+        .Finalize(net.NewOperatorDef());
+  } else {
+    OpDefBuilder("ScalarMath", "ScalarMathBM")
+        .Input("Input")
+        .Output("Output")
+        .AddIntArg("type", type)
+        .AddFloatArg("x", x)
+        .Finalize(net.NewOperatorDef());
+  }
+
+  // Warm-up
+  for (int i = 0; i < 5; ++i) {
+    net.RunOp(D);
+  }
+  net.Sync();
+
+  mace::testing::StartTiming();
+  while (iters--) {
+    net.RunOp(D);
+  }
+  net.Sync();
+}
+
+#define BM_SCALAR_MATH_MACRO(N, C, H, W, X, G, TYPE, DEVICE)              \
+  static void                                                             \
+    BM_SCALAR_MATH_##N##_##C##_##H##_##W##_##X##_##G##_##TYPE##_##DEVICE( \
+      int iters) {                                                   \
+    const int64_t tot = static_cast<int64_t>(iters) * N * C * H * W; \
+    mace::testing::MaccProcessed(tot);                               \
+    mace::testing::BytesProcessed(tot *(sizeof(TYPE)));              \
+    ScalarMath<DEVICE, TYPE>(iters, N, C, H, W, X, G);               \
+  }                                                                  \
+  BENCHMARK(                                                              \
+    BM_SCALAR_MATH_##N##_##C##_##H##_##W##_##X##_##G##_##TYPE##_##DEVICE)
+
+#define BM_SCALAR_MATH(N, C, H, W, X, G)                 \
+  BM_SCALAR_MATH_MACRO(N, C, H, W, X, G, float, CPU);    \
+  BM_SCALAR_MATH_MACRO(N, C, H, W, X, G, float, OPENCL); \
+  BM_SCALAR_MATH_MACRO(N, C, H, W, X, G, half, OPENCL);
+
+BM_SCALAR_MATH(1, 1, 512, 512, 2, 0);
+BM_SCALAR_MATH(1, 3, 128, 128, 2, 1);
+BM_SCALAR_MATH(1, 3, 512, 512, 2, 2);
+BM_SCALAR_MATH(1, 32, 112, 112, 2, 3);
+BM_SCALAR_MATH(1, 64, 256, 256, 3, 0);
+BM_SCALAR_MATH(1, 64, 512, 512, 3, 1);
+BM_SCALAR_MATH(1, 128, 56, 56, 3, 2);
+BM_SCALAR_MATH(1, 128, 256, 256, 3, 3);
+BM_SCALAR_MATH(1, 256, 14, 14, 3, 0);
+BM_SCALAR_MATH(1, 512, 14, 14, 3, 1);
+BM_SCALAR_MATH(1, 1024, 7, 7, 3, 2);
+BM_SCALAR_MATH(32, 1, 256, 256, 3, 3);
+BM_SCALAR_MATH(32, 3, 256, 256, 3, 2);
+
+}  // namespace test
+}  // namespace ops
+}  // namespace mace

mace/ops/scalar_math_test.cc

+//
+// Copyright (c) 2017 XiaoMi All rights reserved.
+//
+
+#include "mace/core/operator.h"
+#include "mace/ops/ops_test_util.h"
+#include "../kernels/scalar_math.h"
+
+namespace mace {
+namespace ops {
+namespace test {
+
+class ScalarMathOpTest : public OpsTestBase {};
+
+
+template <DeviceType D>
+void Simple(const kernels::ScalarMathType type,
+            const std::vector<index_t> &shape,
+            const std::vector<float> &input0,
+            const float x,
+            const std::vector<float> &output) {
+  // Construct graph
+  OpsTestNet net;
+
+  // Add input data
+  net.AddInputFromArray<D, float>("Input1", shape, input0);
+
+  if (D == DeviceType::CPU) {
+    OpDefBuilder("ScalarMath", "ScalarMathTest")
+        .Input("Input1")
+        .AddIntArg("type", static_cast<int>(type))
+        .AddFloatArg("x", x)
+        .Output("Output")
+        .Finalize(net.NewOperatorDef());
+
+    // Run
+    net.RunOp(D);
+  } else {
+    BufferToImage<D, half>(&net, "Input1", "InputImg1",
+                           kernels::BufferType::IN_OUT_CHANNEL);
+    OpDefBuilder("ScalarMath", "ScalarMathTest")
+        .Input("InputImg1")
+        .AddIntArg("type", static_cast<int>(type))
+        .AddFloatArg("x", x)
+        .Output("OutputImg")
+        .Finalize(net.NewOperatorDef());
+
+    // Run
+    net.RunOp(D);
+
+    ImageToBuffer<D, float>(&net, "OutputImg", "Output",
+                            kernels::BufferType::IN_OUT_CHANNEL);
+  }
+
+  auto expected = CreateTensor<float>(shape, output);
+
+  ExpectTensorNear<float>(*expected, *net.GetOutput("Output"), 1e-3);
+}
+
+TEST_F(ScalarMathOpTest, CPUSimple) {
+  Simple<DeviceType::CPU>(kernels::ScalarMathType::MUL, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 0.1, {0.1, 0.2, .3, .4, .5, .6});
+
+  Simple<DeviceType::CPU>(kernels::ScalarMathType::ADD, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 2.0, {3, 4, 5, 6, 7, 8});
+
+  Simple<DeviceType::CPU>(kernels::ScalarMathType::DIV, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 0.1, {10, 20, 30, 40, 50, 60});
+
+  Simple<DeviceType::CPU>(kernels::ScalarMathType::SUB, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 2.0, {-1, 0, 1, 2, 3, 4});
+}
+
+TEST_F(ScalarMathOpTest, GPUSimple) {
+  Simple<DeviceType::OPENCL>(kernels::ScalarMathType::MUL, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 0.1, {0.1, 0.2, .3, .4, .5, .6});
+
+  Simple<DeviceType::OPENCL>(kernels::ScalarMathType::ADD, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 2.0, {3, 4, 5, 6, 7, 8});
+
+  Simple<DeviceType::OPENCL>(kernels::ScalarMathType::DIV, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 0.1, {10, 20, 30, 40, 50, 60});
+
+  Simple<DeviceType::OPENCL>(kernels::ScalarMathType::SUB, {1, 1, 2, 3},
+                          {1, 2, 3, 4, 5, 6}, 2.0, {-1, 0, 1, 2, 3, 4});
+}
+
+template <DeviceType D, typename T>
+void RandomTest(const kernels::ScalarMathType type,
+                const std::vector<index_t> &shape) {
+  testing::internal::LogToStderr();
+  srand(time(NULL));
+
+  // Construct graph
+  OpsTestNet net;
+
+  // Add input data
+  net.AddRandomInput<D, float>("Input1", shape);
+
+  OpDefBuilder("ScalarMath", "ScalarMathTest")
+      .Input("Input1")
+      .AddIntArg("type", static_cast<int>(type))
+      .AddFloatArg("x", 1.2)
+      .Output("Output")
+      .Finalize(net.NewOperatorDef());
+
+  // Run
+  net.RunOp();
+
+  BufferToImage<D, T>(&net, "Input1", "InputImg1",
+                      kernels::BufferType::IN_OUT_CHANNEL);
+
+  OpDefBuilder("ScalarMath", "ScalarMathTest")
+      .Input("InputImg1")
+      .AddIntArg("type", static_cast<int>(type))
+      .AddFloatArg("x", 1.2)
+      .AddIntArg("T", static_cast<int>(DataTypeToEnum<T>::value))
+      .Output("OutputImg")
+      .Finalize(net.NewOperatorDef());
+
+  // Run
+  net.RunOp(D);
+
+  ImageToBuffer<D, float>(&net, "OutputImg", "OPENCLOutput",
+                          kernels::BufferType::IN_OUT_CHANNEL);
+
+  if (DataTypeToEnum<T>::value == DT_FLOAT) {
+    ExpectTensorNear<float>(*net.GetTensor("Output"),
+                            *net.GetOutput("OPENCLOutput"), 1e-3);
+  } else {
+    ExpectTensorNear<float>(*net.GetTensor("Output"),
+                            *net.GetOutput("OPENCLOutput"), 1e-1);
+  }
+}
+
+TEST_F(ScalarMathOpTest, OPENCLRandomFloat) {
+  RandomTest<DeviceType::OPENCL, float>(kernels::ScalarMathType::MUL,
+                                        {3, 23, 37, 19});
+  RandomTest<DeviceType::OPENCL, float>(kernels::ScalarMathType::ADD,
+                                        {13, 32, 32, 64});
+  RandomTest<DeviceType::OPENCL, float>(kernels::ScalarMathType::SUB,
+                                        {3, 32, 32, 64});
+  RandomTest<DeviceType::OPENCL, float>(kernels::ScalarMathType::DIV,
+                                        {13, 32, 32, 64});
+}
+
+TEST_F(ScalarMathOpTest, OPENCLRandomHalf) {
+  RandomTest<DeviceType::OPENCL, half>(kernels::ScalarMathType::MUL,
+                                       {3, 23, 37, 19});
+  RandomTest<DeviceType::OPENCL, half>(kernels::ScalarMathType::ADD,
+                                       {13, 32, 32, 64});
+  RandomTest<DeviceType::OPENCL, half>(kernels::ScalarMathType::SUB,
+                                       {3, 32, 32, 64});
+  RandomTest<DeviceType::OPENCL, half>(kernels::ScalarMathType::DIV,
+                                       {13, 32, 32, 64});
+}
+
+}  // namespace test
+}  // namespace ops
+}  // namespace mace

mace/python/tools/tf_converter_lib.py

@@ -19,6 +19,16 @@ pooling_type_mode = {
   'MaxPool': 2
 }
 
+# the order should be the same as eltwise type's order
+math_type_mode = {
+  'MUL': 0,
+  'ADD': 1,
+  'MAX': 2,
+  'MIN': 3,
+  'SUB': 4,
+  'DIV': 5
+}
+
 buffer_type_map = {
   'CONV2D_FILTER' : 0,
   'IN_OUT_CHANNEL' : 1,
@@ -623,6 +633,64 @@ class TFConverter(object):
     self.resolved_ops[op.name] = 1
     self.unused_tensor.add(get_input_tensor(op, 1).name)
 
+  def convert_neg(self, op):
+    op_def = self.net_def.op.add()
+    arg = op_def.arg.add()
+    arg.name = 'T'
+    arg.i = self.dt
+    op_def.name = op.name
+    op_def.type = "Neg"
+    op_def.input.extend([input.name for input in op.inputs])
+    op_def.output.extend([output.name for output in op.outputs])
+    self.add_output_shape(op.outputs, op_def)
+    self.resolved_ops[op.name] = 1
+    
+  def convert_math(self, op, math_type):
+    op_def = self.net_def.op.add()
+    arg = op_def.arg.add()
+    arg.name = 'T'
+    arg.i = self.dt
+    op_def.name = op.name
+    input_tensor0 = get_input_tensor(op, 0)
+    input_tensor1 = get_input_tensor(op, 1)
+    
+    if input_tensor0.shape == input_tensor1.shape:
+      op_def.type = "Eltwise"
+      op_def.input.extend([input.name for input in op.inputs])
+    else:
+      op_def.type = "ScalarMath"
+      x_value = 0
+      if len(input_tensor1.shape)==4:
+        op_def.input.extend([op.inputs[1].name])
+        x_value = get_input_tensor(op, 0).eval().astype(np.float32)
+      else:
+        op_def.input.extend([op.inputs[0].name])
+        x_value = get_input_tensor(op, 1).eval().astype(np.float32)
+      x_arg = op_def.arg.add()
+      x_arg.name = 'x'
+      x_arg.f = x_value
+    type_arg = op_def.arg.add()
+    type_arg.name = 'type'
+    type_arg.i = math_type_mode[math_type]
+    op_def.output.extend([output.name for output in op.outputs])
+    self.add_output_shape(op.outputs, op_def)
+    self.resolved_ops[op.name] = 1
+
+  def convert_depth_to_space(self, op, d2s):
+    op_def = self.net_def.op.add()
+    arg = op_def.arg.add()
+    arg.name = 'T'
+    arg.i = self.dt
+    op_def.name = op.name
+    op_def.type = op.type
+    op_def.input.extend([op.inputs[0].name])
+    op_def.output.extend([output.name for output in op.outputs])
+    size_arg = op_def.arg.add()
+    size_arg.name = 'block_size'
+    size_arg.i = op.get_attr('block_size')
+    self.add_output_shape(op.outputs, op_def)
+    self.resolved_ops[op.name] = 1
+
   def convert_bias_add(self, op):
     op_def = mace_pb2.OperatorDef()
     arg = op_def.arg.add()
@@ -850,6 +918,16 @@ class TFConverter(object):
         self.convert_space_to_batch(op, False)
       elif op.type == 'BatchToSpaceND':
         self.convert_space_to_batch(op, True)
+      elif op.type == 'DepthToSpace':
+		self.convert_depth_to_space(op, True)
+      elif op.type == 'SpaceToDepth':
+		self.convert_depth_to_space(op, False)
+      elif op.type == 'Neg':
+		self.convert_neg(op)
+      elif op.type == 'Mul':
+		self.convert_math(op, 'MUL')
+      elif op.type == 'Sub':
+		self.convert_math(op, 'SUB')
       elif self.is_softmax(op):
         self.convert_softmax(op)
       elif op.type in ['Relu', 'Sigmoid', 'Tanh']: