提交 a3b45b71 编写于 作者: C chenzupeng

fix bug in leakyrelu

上级 77dd91a6
......@@ -5,33 +5,18 @@
#define MIN(X, Y) (X < Y ? X : Y)
__constant sampler_t smp_zero = CLK_NORMALIZED_COORDS_FALSE | CLK_ADDRESS_CLAMP | CLK_FILTER_NEAREST;
__kernel void LeakyRelu_NHWC4(__read_only image2d_t input, __write_only image2d_t output, const int4 img_shape,
__global FLT4 *alpha, const int4 input_shape) {
__kernel void LeakyRelu(__read_only image2d_t input, __write_only image2d_t output, const int4 img_shape,
const float alpha) {
int Y = get_global_id(0); // H
int X = get_global_id(1); // W C4
if (X >= img_shape.z || Y >= img_shape.y) return;
int C = X % UP_DIV(input_shape.w, SLICES);
FLT4 in_c4 = READ_IMAGE(input, smp_zero, (int2)(X, Y));
FLT4 tmp;
tmp.x = in_c4.x > 0.0f ? in_c4.x : in_c4.x * alpha[C].x;
tmp.y = in_c4.y > 0.0f ? in_c4.y : in_c4.y * alpha[C].y;
tmp.z = in_c4.z > 0.0f ? in_c4.z : in_c4.z * alpha[C].z;
tmp.w = in_c4.w > 0.0f ? in_c4.w : in_c4.w * alpha[C].w;
WRITE_IMAGE(output, (int2)(X, Y), tmp);
}
__kernel void LeakyRelu_NC4HW4(__read_only image2d_t input, __write_only image2d_t output, const int4 img_shape,
__global FLT4 *alpha, const int4 input_shape) {
int Y = get_global_id(0); // C4 H
int X = get_global_id(1); // W
if (X >= img_shape.z || Y >= img_shape.y) return;
int C = Y / input_shape.y;
FLT4 in_c4 = READ_IMAGE(input, smp_zero, (int2)(X, Y));
FLT4 tmp;
tmp.x = in_c4.x > 0.0f ? in_c4.x : in_c4.x * alpha[C].x;
tmp.y = in_c4.y > 0.0f ? in_c4.y : in_c4.y * alpha[C].y;
tmp.z = in_c4.z > 0.0f ? in_c4.z : in_c4.z * alpha[C].z;
tmp.w = in_c4.w > 0.0f ? in_c4.w : in_c4.w * alpha[C].w;
FLT alpha_f = TO_FLT(alpha);
tmp.x = in_c4.x > 0.0f ? in_c4.x : in_c4.x * alpha_f;
tmp.y = in_c4.y > 0.0f ? in_c4.y : in_c4.y * alpha_f;
tmp.z = in_c4.z > 0.0f ? in_c4.z : in_c4.z * alpha_f;
tmp.w = in_c4.w > 0.0f ? in_c4.w : in_c4.w * alpha_f;
WRITE_IMAGE(output, (int2)(X, Y), tmp);
}
......
......@@ -39,46 +39,7 @@ using mindspore::schema::PrimitiveType_Activation;
namespace mindspore::kernel {
void ActivationOpenClKernel::InitBuffer() {
auto allocator = lite::opencl::OpenCLRuntime::GetInstance()->GetAllocator();
int elem_num = UP_ROUND(nhwc_shape_[3], C4NUM);
alpha_buff_ = allocator->Malloc(elem_num * fp_size);
alpha_buff_ = allocator->MapBuffer(alpha_buff_, CL_MAP_WRITE, nullptr, true);
memset(alpha_buff_, 0x00, elem_num * fp_size);
if (in_tensors_.size() == 1) {
if (enable_fp16_) {
uint16_t alpha_fp16 = Float32ToShort(alpha_);
auto alpha_buff_fp16 = reinterpret_cast<uint16_t *>(alpha_buff_);
for (int i = 0; i < nhwc_shape_[3]; i++) {
alpha_buff_fp16[i] = alpha_fp16;
}
} else {
auto alpha_buff_fp16 = reinterpret_cast<float *>(alpha_buff_);
for (int i = 0; i < nhwc_shape_[3]; i++) {
alpha_buff_fp16[i] = alpha_;
}
}
} else {
if (enable_fp16_) {
if (in_tensors_[1]->data_type() == kNumberTypeFloat32) {
auto alpha_buff_fp16 = reinterpret_cast<uint16_t *>(alpha_buff_);
for (int i = 0; i < nhwc_shape_[3]; i++) {
alpha_buff_fp16[i] = Float32ToShort(reinterpret_cast<float *>(in_tensors_[0]->Data())[i]);
}
} else {
memcpy(alpha_buff_, in_tensors_[0]->Data(), nhwc_shape_[3] * fp_size);
}
} else {
if (in_tensors_[1]->data_type() == kNumberTypeFloat16) {
MS_LOG(WARNING) << "fp16 model run in fp32 mode not support.";
memcpy(alpha_buff_, in_tensors_[0]->Data(), nhwc_shape_[3] * fp_size);
} else {
memcpy(alpha_buff_, in_tensors_[0]->Data(), nhwc_shape_[3] * fp_size);
}
}
}
allocator->UnmapBuffer(alpha_buff_);
}
void ActivationOpenClKernel::InitBuffer() {}
int ActivationOpenClKernel::Init() {
in_size_ = in_tensors_[0]->shape().size();
......@@ -102,9 +63,6 @@ int ActivationOpenClKernel::Init() {
MS_LOG(ERROR) << "Activate fun only support dim=4 or 2, but your dim=" << in_size_;
return RET_ERROR;
}
if (type_ == ActivationType_LEAKY_RELU) {
InitBuffer();
}
std::map<int, std::vector<std::string>> Program_Kernel{
{ActivationType_LEAKY_RELU, std::vector<std::string>{"LEAKY_RELU", "LeakyRelu"}},
{ActivationType_RELU, std::vector<std::string>{"RELU", "Relu"}},
......@@ -119,9 +77,6 @@ int ActivationOpenClKernel::Init() {
std::set<std::string> build_options;
ocl_runtime->LoadSource(Program_Kernel[type_][0], source);
std::string kernel_name = Program_Kernel[type_][1];
if (type_ == ActivationType_LEAKY_RELU) {
kernel_name += "_" + std::string(EnumNameFormat(op_format_));
}
ocl_runtime->BuildKernel(kernel_, Program_Kernel[type_][0], kernel_name, build_options);
in_ori_format_ = in_tensors_[0]->GetFormat();
out_ori_format_ = out_tensors_[0]->GetFormat();
......@@ -140,10 +95,7 @@ int ActivationOpenClKernel::Run() {
ocl_runtime->SetKernelArg(kernel_, arg_idx++, out_tensors_[0]->Data());
ocl_runtime->SetKernelArg(kernel_, arg_idx++, img2d_shape);
if (type_ == ActivationType_LEAKY_RELU) {
ocl_runtime->SetKernelArg(kernel_, arg_idx++, alpha_buff_, lite::opencl::MemType::BUF);
cl_int4 input_shape = {static_cast<int>(nhwc_shape_[0]), static_cast<int>(nhwc_shape_[1]),
static_cast<int>(nhwc_shape_[2]), static_cast<int>(nhwc_shape_[3])};
ocl_runtime->SetKernelArg(kernel_, arg_idx++, input_shape);
ocl_runtime->SetKernelArg(kernel_, arg_idx++, alpha_);
}
std::vector<size_t> local = {};
std::vector<size_t> global = {static_cast<size_t>(img2d_shape.s[1]), static_cast<size_t>(img2d_shape.s[2])};
......
......@@ -22,6 +22,7 @@
#include "src/kernel_registry.h"
#include "include/errorcode.h"
#include "nnacl/fp32/common_func.h"
#include "src/runtime/kernel/opencl/kernel/prelu.h"
#include "src/runtime/opencl/opencl_runtime.h"
#include "src/runtime/kernel/opencl/cl/prelu.cl.inc"
......@@ -35,18 +36,38 @@ using mindspore::schema::PrimitiveType_PReLU;
namespace mindspore::kernel {
void PReluOpenCLKernel::InitBuffer() {
int C = in_tensors_[1]->shape()[0];
int div_ci = UP_DIV(C, C4NUM);
auto allocator = lite::opencl::OpenCLRuntime::GetInstance()->GetAllocator();
int elem_num = in_tensors_[0]->shape().size() == 2 ? in_tensors_[0]->shape()[1] : in_tensors_[0]->shape()[3];
int elem_num_c4 = UP_DIV(elem_num, C4NUM);
size_t img_dtype = CL_FLOAT;
if (enable_fp16_) {
img_dtype = CL_HALF_FLOAT;
}
std::vector<size_t> img_size{size_t(div_ci), 1, img_dtype};
PReluWeight_ = allocator->Malloc(div_ci * C4NUM * fp_size, img_size);
std::vector<size_t> img_size{size_t(elem_num_c4), 1, img_dtype};
PReluWeight_ = allocator->Malloc(elem_num_c4 * C4NUM * fp_size, img_size);
PReluWeight_ = allocator->MapBuffer(PReluWeight_, CL_MAP_WRITE, nullptr, true);
memset(PReluWeight_, 0x00, div_ci * C4NUM * fp_size);
memcpy(PReluWeight_, in_tensors_[1]->Data(), C * fp_size);
memset(PReluWeight_, 0x00, elem_num_c4 * C4NUM * fp_size);
if (enable_fp16_) {
if (in_tensors_[1]->data_type() == kNumberTypeFloat32) {
auto PReluWeight_fp16 = reinterpret_cast<uint16_t *>(PReluWeight_);
auto in_tensor_data_fp32 = reinterpret_cast<float *>(in_tensors_[1]->Data());
for (int i = 0; i < elem_num; i++) {
PReluWeight_fp16[i] = Float32ToShort(in_tensor_data_fp32[i]);
}
} else {
memcpy(PReluWeight_, in_tensors_[1]->Data(), elem_num * fp_size);
}
} else {
if (in_tensors_[1]->data_type() == kNumberTypeFloat16) {
auto PReluWeight_fp32 = reinterpret_cast<float *>(PReluWeight_);
auto in_tensor_data_fp16 = reinterpret_cast<uint16_t *>(in_tensors_[1]->Data());
for (int i = 0; i < elem_num; i++) {
PReluWeight_fp32[i] = ShortToFloat32(in_tensor_data_fp16[i]);
}
} else {
memcpy(PReluWeight_, in_tensors_[1]->Data(), elem_num * fp_size);
}
}
allocator->UnmapBuffer(PReluWeight_);
}
......
......@@ -432,7 +432,7 @@ TEST_F(TestActivationOpenCL, LeakyReluFp_dim4) {
std::vector<int> input_shape = {1, 9}; // need modify
auto tensor_type = schema::NodeType_ValueNode;
schema::Format format = schema::Format_NC; // need modify
schema::Format op_format = schema::Format_NC4; // need modify
schema::Format op_format = schema::Format_NHWC4; // need modify
auto *input_tensor = new (std::nothrow) lite::tensor::Tensor(data_type, input_shape, format, tensor_type);
if (input_tensor == nullptr) {
MS_LOG(ERROR) << "new input tensor error!";
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册