From b4815227b7d2282c7a397c7cadb7bc9f3483480d Mon Sep 17 00:00:00 2001
From: lltcggie <lltcggie@gmail.com>
Date: Sat, 6 Jun 2015 21:40:55 +0900
Subject: [PATCH] =?UTF-8?q?=E5=88=86=E5=89=B2=E3=83=96=E3=83=AD=E3=83=83?=
 =?UTF-8?q?=E3=82=AF=E3=82=92=E4=B8=A6=E5=88=97=E3=81=A7=E5=A4=89=E6=8F=9B?=
 =?UTF-8?q?=E3=81=99=E3=82=8B=E3=82=88=E3=81=86=E3=81=AB=E3=81=97=E3=81=9F?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 common/tinypl.hpp                             | 235 ++++++++++++++++++
 common/waifu2x.cpp                            | 167 +++++++++----
 common/waifu2x.h                              |  28 ++-
 waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj   |   5 +-
 .../waifu2x-caffe-gui.vcxproj.filters         |   3 +
 waifu2x-caffe/waifu2x-caffe.vcxproj           |   5 +-
 waifu2x-caffe/waifu2x-caffe.vcxproj.filters   |   3 +
 7 files changed, 388 insertions(+), 58 deletions(-)
 create mode 100644 common/tinypl.hpp

diff --git a/common/tinypl.hpp b/common/tinypl.hpp
new file mode 100644
index 0000000..8102b8f
--- /dev/null
+++ b/common/tinypl.hpp
@@ -0,0 +1,235 @@
+/**
+ * Tiny Parallel Library
+ *
+ * The library implements following parallel algorithms that its
+ * interface are compatible with Intel TBB and Microsoft PPL.
+ *  - parallel_for_each(first,last,func)
+ *  - parallel_for(first,last,func)
+ *  - parallel_invoke(f1,f2,...)  (up to 4 args)
+ */
+#ifndef TINYPL_HPP
+#define TINYPL_HPP
+
+#include <algorithm>
+#include <iterator>
+#include <memory>
+#include <thread>
+#include <atomic>
+#include <functional>
+//#define BOOST_ASIO_ENABLE_HANDLER_TRACKING
+#include <boost/asio.hpp>
+
+#ifndef TINYPL_WORKERNUM
+#define TINYPL_WORKERNUM 0
+#endif
+
+#ifndef TINYPL_MIN_ITERATE
+#define TINYPL_MIN_ITERATE 1
+#endif
+
+
+namespace tinypl
+{
+	namespace impl {
+
+		// task scheduler
+		class scheduler {
+			friend struct waiter;
+		public:
+			explicit scheduler(std::size_t thnum = 0)
+				: worker_(new boost::asio::io_service::work(iosrv_))
+			{
+				if (thnum == 0)
+					thnum = std::max(1u, std::thread::hardware_concurrency());
+				thnum_ = thnum;
+				// start worker threads
+				for (std::size_t i = 0; i < thnum - 1; ++i)
+					thpool_.emplace_back(std::bind(static_cast<std::size_t(boost::asio::io_service::*)(void)>(&boost::asio::io_service::run), &iosrv_));
+			}
+
+			~scheduler()
+			{
+				// stop all worker threads
+				worker_.reset();
+				for (auto &th : thpool_)
+					th.join();
+			}
+
+			// # of worker threads
+			std::size_t worker_num() const { return thnum_; }
+
+			// euqueue task
+			template <class F>
+			void enqueue(F f) { iosrv_.post(f); }
+
+			std::vector<std::thread::id> get_thread_pool_id_list() const
+			{
+				std::vector<std::thread::id> list;
+				list.reserve(thpool_.size());
+				for (const auto &th : thpool_)
+					list.push_back(th.get_id());
+
+				return list;
+			}
+
+		public:
+			// get scheduler object
+			static scheduler& instance()
+			{
+				static scheduler sched(TINYPL_WORKERNUM);
+				return sched;
+			}
+
+		private:
+			boost::asio::io_service iosrv_;
+			std::unique_ptr<boost::asio::io_service::work> worker_;
+			std::vector<std::thread> thpool_;
+			std::size_t thnum_;
+		};
+
+		// task waiter
+		struct waiter {
+			scheduler& sched_;
+			volatile std::atomic<uint32_t> count_;
+
+			waiter(scheduler& sched, unsigned int count)
+				: sched_(sched), count_(count) {}
+			~waiter()
+			{
+				while (0 < count_) {
+					sched_.iosrv_.poll_one();
+					// FIXME: It may cause heavy busyloop in worst-case scenario.
+				}
+			}
+
+			struct holder {
+				explicit holder(waiter& w) : w_(w) {}
+				~holder() { w_.count_--; }
+				waiter& w_;
+			};
+		};
+
+
+		// task of parallel_for_each algorithm
+		template <class Itr, class Func>
+		void parallel_foreach_task(waiter* w, Itr first, Itr last, const Func& func)
+		{
+			waiter::holder h(*w);
+			while (first != last)
+				func(*first++);
+		}
+
+		// task of parallel_for algorithm
+		template <class IdxType, class Func>
+		void parallel_for_task(waiter* w, IdxType first, IdxType last, const Func& func)
+		{
+			waiter::holder h(*w);
+			while (first < last)
+				func(first++);
+		}
+
+		// task of parallel_invoke algorithm
+		template <class Func>
+		void parallel_invoke_task(waiter* w, const Func& func)
+		{
+			waiter::holder h(*w);
+			func();
+		}
+
+	} // namespace impl
+
+
+	/**
+	 * parallel_for_each algorithm
+	 */
+	template <class Itr, class Func>
+	void parallel_for_each(Itr first, Itr last, const Func& func)
+	{
+		impl::scheduler& sched = impl::scheduler::instance();
+		std::size_t range = std::distance(first, last);
+		std::size_t block = std::max(range / sched.worker_num(), std::size_t(TINYPL_MIN_ITERATE));
+		impl::waiter w(sched, (range + block - 1) / block);
+		for (Itr next = first; first != last; first = next) {
+			std::advance(next, std::min(range, block));
+			range -= std::min(range, block);
+			if (next != last) {
+				sched.enqueue(boost::bind(&impl::parallel_foreach_task<Itr, Func>, &w, first, next, func));
+			}
+			else {
+				impl::parallel_foreach_task<Itr, Func>(&w, first, next, func);
+			}
+		}
+	}
+
+	/**
+	* parallel_for algorithm
+	*/
+	template <class IdxType, class Func>
+	void parallel_for(impl::scheduler& sched, IdxType first, IdxType last, const Func& func)
+	{
+		IdxType range = last - first;
+		IdxType block = static_cast<IdxType>(std::max(range / sched.worker_num(), std::size_t(TINYPL_MIN_ITERATE)));
+		impl::waiter w(sched, (range + block - 1) / block);
+		for (IdxType next = first; first < last; first = next) {
+			next = std::min(last, next + block);
+			if (next < last) {
+				sched.enqueue(std::bind(&impl::parallel_for_task<IdxType, Func>, &w, first, next, func));
+			}
+			else {
+				impl::parallel_for_task<IdxType, Func>(&w, first, next, func);
+			}
+		}
+	}
+
+	/**
+	* parallel_for algorithm
+	*/
+	template <class IdxType, class Func>
+	void parallel_for(IdxType first, IdxType last, const Func& func)
+	{
+		impl::scheduler& sched = impl::scheduler::instance();
+		parallel_for(sched, fist, last, func);
+	}
+
+	/**
+	 * parallel_invoke algorithm (2 args)
+	 */
+	template <class F1, class F2>
+	void parallel_invoke(const F1& f1, const F2& f2)
+	{
+		impl::scheduler& sched = impl::scheduler::instance();
+		impl::waiter w(sched, 1);
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F1>, &w, f1));
+		f2();
+	}
+
+	/**
+	 * parallel_invoke algorithm (3 args)
+	 */
+	template <class F1, class F2, class F3>
+	void parallel_invoke(const F1& f1, const F2& f2, const F3& f3)
+	{
+		impl::scheduler& sched = impl::scheduler::instance();
+		impl::waiter w(sched, 2);
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F1>, &w, f1));
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F2>, &w, f2));
+		f3();
+	}
+
+	/**
+	 * parallel_invoke algorithm (4 args)
+	 */
+	template <class F1, class F2, class F3, class F4>
+	void parallel_invoke(const F1& f1, const F2& f2, const F3& f3, const F4& f4)
+	{
+		impl::scheduler& sched = impl::scheduler::instance();
+		impl::waiter w(sched, 3);
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F1>, &w, f1));
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F2>, &w, f2));
+		sched.enqueue(boost::bind(&impl::parallel_invoke_task<F3>, &w, f3));
+		f4();
+	}
+
+} // namespace tinypl
+
+#endif
diff --git a/common/waifu2x.cpp b/common/waifu2x.cpp
index 2a243e9..49de440 100644
--- a/common/waifu2x.cpp
+++ b/common/waifu2x.cpp
@@ -9,6 +9,7 @@
 #include <boost/algorithm/string.hpp>
 #include <chrono>
 #include <cuda_runtime.h>
+#include "tinypl.hpp"
 
 #if defined(WIN32) || defined(WIN64)
 #include <Windows.h>
@@ -62,7 +63,7 @@ static std::once_flag waifu2x_cudnn_once_flag;
 		} \
 	} while (0)
 
-Waifu2x::Waifu2x() : is_inited(false), isCuda(false), block(nullptr), dummy_data(nullptr), out_block(nullptr)
+Waifu2x::Waifu2x() : job(1), is_inited(false), isCuda(false)
 {
 }
 
@@ -375,7 +376,7 @@ Waifu2x::eWaifu2xError Waifu2x::ConstractNet(boost::shared_ptr<caffe::Net<float>
 }
 
 // ネットワークを使って画像を再構築する
-Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(boost::shared_ptr<caffe::Net<float>> net, cv::Mat &im)
+Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(std::vector<boost::shared_ptr<caffe::Net<float>>> nets, cv::Mat &im)
 {
 	const auto Height = im.size().height;
 	const auto Width = im.size().width;
@@ -390,18 +391,6 @@ Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(boost::shared_ptr<caffe::Net<fl
 
 	try
 	{
-		const auto input_layer =
-			boost::dynamic_pointer_cast<caffe::MemoryDataLayer<float>>(
-			net->layer_by_name("image_input_layer"));
-		assert(input_layer);
-
-		const auto conv7_layer =
-			boost::dynamic_pointer_cast<caffe::ConvolutionLayer<float>>(
-			net->layer_by_name("conv7_layer"));
-		assert(conv7_layer);
-
-		input_layer->set_batch_size(batch_size);
-
 		const int WidthNum = Width / output_size;
 		const int HeightNum = Height / output_size;
 
@@ -410,9 +399,41 @@ Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(boost::shared_ptr<caffe::Net<fl
 		const int input_block_plane_size = block_size * block_size;
 		const int output_block_plane_size = crop_size * crop_size;
 
-		// 画像は(消費メモリの都合上)output_size*output_sizeに分けて再構築する
-		for (int num = 0; num < BlockNum; num += batch_size)
+		const int BatchNum = BlockNum / batch_size + (BlockNum % batch_size != 0 ? 1 : 0);
+
+		for (auto net : nets)
 		{
+			const auto input_layer =
+				boost::dynamic_pointer_cast<caffe::MemoryDataLayer<float>>(
+				net->layer_by_name("image_input_layer"));
+			assert(input_layer);
+
+			input_layer->set_batch_size(batch_size);
+		}
+
+		// 画像は(消費メモリの都合上)output_size*output_sizeに分けて再構築する
+		tinypl::parallel_for(*net_scheduler, 0, BatchNum, [&](const int batch_n)
+		{
+			const auto id = std::this_thread::get_id();
+			const auto net_scheduler_id_map_it = net_scheduler_id_map.find(id);
+
+			assert(net_scheduler_id_map_it != net_scheduler_id_map.end());
+
+			const int index = net_scheduler_id_map_it->second;
+
+			auto net = nets[index];
+
+			float *block = blocks[index];
+			float *dummy_data = dummy_datas[index];
+			float *out_block = out_blocks[index];
+
+			const auto input_layer =
+				boost::dynamic_pointer_cast<caffe::MemoryDataLayer<float>>(
+				net->layer_by_name("image_input_layer"));
+			assert(input_layer);
+
+			const int num = batch_n * batch_size;
+
 			const int processNum = (BlockNum - num) >= batch_size ? batch_size : BlockNum - num;
 
 			if (processNum < batch_size)
@@ -455,7 +476,7 @@ Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(boost::shared_ptr<caffe::Net<fl
 			}
 
 			// ネットワークに画像を入力
-			input_layer->Reset(block, dummy_data, input_block_plane_size * batch_size);
+			input_layer->Reset(block, dummy_data, input_block_plane_size * processNum);
 
 			// 計算
 			auto out = net->ForwardPrefilled(nullptr);
@@ -487,7 +508,7 @@ Waifu2x::eWaifu2xError Waifu2x::ReconstructImage(boost::shared_ptr<caffe::Net<fl
 				for (int i = 0; i < crop_size; i++)
 					memcpy(imptr + (h + i) * Line + w, fptr + i * crop_size, crop_size * sizeof(float));
 			}
-		}
+		});
 	}
 	catch (...)
 	{
@@ -519,6 +540,8 @@ Waifu2x::eWaifu2xError Waifu2x::init(int argc, char** argv, const std::string &M
 		crop_size = CropSize;
 		batch_size = BatchSize;
 
+		job = 2;
+
 		output_size = crop_size - offset * 2;
 		block_size = crop_size + layer_num * 2;
 		original_width_height = 128 + layer_num * 2;
@@ -577,13 +600,18 @@ Waifu2x::eWaifu2xError Waifu2x::init(int argc, char** argv, const std::string &M
 			const std::string model_path = (mode_dir_path / "srcnn.prototxt").string();
 			const std::string param_path = (mode_dir_path / ("noise" + std::to_string(noise_level) + "_model.json")).string();
 
-			ret = ConstractNet(net_noise, model_path, process);
-			if (ret != eWaifu2xError_OK)
-				return ret;
+			net_noises.resize(job);
 
-			ret = LoadParameter(net_noise, param_path);
-			if (ret != eWaifu2xError_OK)
-				return ret;
+			for (auto &net_noise : net_noises)
+			{
+				ret = ConstractNet(net_noise, model_path, process);
+				if (ret != eWaifu2xError_OK)
+					return ret;
+
+				ret = LoadParameter(net_noise, param_path);
+				if (ret != eWaifu2xError_OK)
+					return ret;
+			}
 		}
 
 		if (mode == "scale" || mode == "noise_scale" || mode == "auto_scale")
@@ -591,33 +619,62 @@ Waifu2x::eWaifu2xError Waifu2x::init(int argc, char** argv, const std::string &M
 			const std::string model_path = (mode_dir_path / "srcnn.prototxt").string();
 			const std::string param_path = (mode_dir_path / "scale2.0x_model.json").string();
 
-			ret = ConstractNet(net_scale, model_path, process);
-			if (ret != eWaifu2xError_OK)
-				return ret;
+			net_scales.resize(job);
 
-			ret = LoadParameter(net_scale, param_path);
-			if (ret != eWaifu2xError_OK)
-				return ret;
+			for (auto &net_scale : net_scales)
+			{
+				ret = ConstractNet(net_scale, model_path, process);
+				if (ret != eWaifu2xError_OK)
+					return ret;
+
+				ret = LoadParameter(net_scale, param_path);
+				if (ret != eWaifu2xError_OK)
+					return ret;
+			}
 		}
 
 		const int input_block_plane_size = block_size * block_size;
 		const int output_block_plane_size = crop_size * crop_size;
 
+		blocks.resize(job);
+		dummy_datas.resize(job);
+		out_blocks.resize(job);
+
 		if (isCuda)
 		{
-			CUDA_CHECK_WAIFU2X(cudaHostAlloc(&block, sizeof(float) * input_block_plane_size * batch_size, cudaHostAllocWriteCombined));
-			CUDA_CHECK_WAIFU2X(cudaHostAlloc(&dummy_data, sizeof(float) * input_block_plane_size * batch_size, cudaHostAllocWriteCombined));
-			CUDA_CHECK_WAIFU2X(cudaHostAlloc(&out_block, sizeof(float) * output_block_plane_size * batch_size, cudaHostAllocDefault));
+			for (auto &block : blocks) {
+				CUDA_CHECK_WAIFU2X(cudaHostAlloc(&block, sizeof(float) * input_block_plane_size * batch_size, cudaHostAllocWriteCombined));
+			}
+			for (auto &dummy_data : dummy_datas) {
+				CUDA_CHECK_WAIFU2X(cudaHostAlloc(&dummy_data, sizeof(float) * input_block_plane_size * batch_size, cudaHostAllocWriteCombined));
+			}
+			for (auto &out_block : out_blocks) {
+				CUDA_CHECK_WAIFU2X(cudaHostAlloc(&out_block, sizeof(float) * output_block_plane_size * batch_size, cudaHostAllocDefault));
+			}
 		}
 		else
 		{
-			block = new float[input_block_plane_size * batch_size];
-			dummy_data = new float[input_block_plane_size * batch_size];
-			out_block = new float[output_block_plane_size * batch_size];
+			for (auto &block : blocks)
+				block = new float[input_block_plane_size * batch_size];
+			for (auto &dummy_data : dummy_datas)
+				dummy_data = new float[input_block_plane_size * batch_size];
+			for (auto &out_block : out_blocks)
+				out_block = new float[output_block_plane_size * batch_size];
 		}
 
-		for (size_t i = 0; i < input_block_plane_size * batch_size; i++)
-			dummy_data[i] = 0.0f;
+		for (auto dummy_data : dummy_datas)
+		{
+			for (size_t i = 0; i < input_block_plane_size * batch_size; i++)
+				dummy_data[i] = 0.0f;
+		}
+
+		net_scheduler.reset(new tinypl::impl::scheduler(job));
+
+		const auto list = net_scheduler->get_thread_pool_id_list();
+		for (size_t i = 0; i < list.size(); i++)
+			net_scheduler_id_map.emplace(list[i], i);
+
+		net_scheduler_id_map.emplace(std::this_thread::get_id(), list.size());
 
 		is_inited = true;
 	}
@@ -631,20 +688,34 @@ Waifu2x::eWaifu2xError Waifu2x::init(int argc, char** argv, const std::string &M
 
 void Waifu2x::destroy()
 {
-	net_noise.reset();
-	net_scale.reset();
+	net_scheduler.reset();
+
+	net_noises.clear();
+	net_scales.clear();
 
 	if (isCuda)
 	{
-		CUDA_HOST_SAFE_FREE(block);
-		CUDA_HOST_SAFE_FREE(dummy_data);
-		CUDA_HOST_SAFE_FREE(out_block);
+		for (auto &block : blocks) {
+			CUDA_HOST_SAFE_FREE(block);
+		}
+		for (auto &dummy_data : dummy_datas) {
+			CUDA_HOST_SAFE_FREE(dummy_data);
+		}
+		for (auto &out_block : out_blocks) {
+			CUDA_HOST_SAFE_FREE(out_block);
+		}
 	}
 	else
 	{
-		SAFE_DELETE_WAIFU2X(block);
-		SAFE_DELETE_WAIFU2X(dummy_data);
-		SAFE_DELETE_WAIFU2X(out_block);
+		for (auto &block : blocks) {
+			SAFE_DELETE_WAIFU2X(block);
+		}
+		for (auto &dummy_data : dummy_datas) {
+			SAFE_DELETE_WAIFU2X(dummy_data);
+		}
+		for (auto &out_block : out_blocks) {
+			SAFE_DELETE_WAIFU2X(out_block);
+		}
 	}
 
 	is_inited = false;
@@ -680,7 +751,7 @@ Waifu2x::eWaifu2xError Waifu2x::waifu2x(const std::string &input_file, const std
 	{
 		PaddingImage(im, im);
 
-		ret = ReconstructImage(net_noise, im);
+		ret = ReconstructImage(net_noises, im);
 		if (ret != eWaifu2xError_OK)
 			return ret;
 
@@ -701,7 +772,7 @@ Waifu2x::eWaifu2xError Waifu2x::waifu2x(const std::string &input_file, const std
 		{
 			Zoom2xAndPaddingImage(im, im, image_size);
 
-			ret = ReconstructImage(net_scale, im);
+			ret = ReconstructImage(net_scales, im);
 			if (ret != eWaifu2xError_OK)
 				return ret;
 
diff --git a/common/waifu2x.h b/common/waifu2x.h
index b35293c..4ec4c31 100644
--- a/common/waifu2x.h
+++ b/common/waifu2x.h
@@ -5,8 +5,11 @@
 #include <vector>
 #include <utility>
 #include <functional>
+#include <memory>
+#include <thread>
 #include <boost/shared_ptr.hpp>
 #include <opencv2/opencv.hpp>
+#include <unordered_map>
 
 
 namespace caffe
@@ -15,6 +18,15 @@ namespace caffe
 	class Net;
 };
 
+namespace tinypl
+{
+	namespace impl
+	{
+		// task scheduler
+		class scheduler;
+	}
+}
+
 class Waifu2x
 {
 public:
@@ -64,15 +76,19 @@ private:
 	double scale_ratio;
 	std::string model_dir;
 	std::string process;
+	int job;
 
 	bool isCuda;
 
-	boost::shared_ptr<caffe::Net<float>> net_noise;
-	boost::shared_ptr<caffe::Net<float>> net_scale;
+	std::vector<boost::shared_ptr<caffe::Net<float>>> net_noises;
+	std::vector<boost::shared_ptr<caffe::Net<float>>> net_scales;
 
-	float *block;
-	float *dummy_data;
-	float *out_block;
+	std::vector<float *> blocks;
+	std::vector<float *> dummy_datas;
+	std::vector<float *> out_blocks;
+
+	std::unique_ptr<tinypl::impl::scheduler> net_scheduler;
+	std::unordered_map<std::thread::id, size_t> net_scheduler_id_map;
 
 private:
 	eWaifu2xError LoadImage(cv::Mat &float_image, const std::string &input_file);
@@ -82,7 +98,7 @@ private:
 	eWaifu2xError CreateZoomColorImage(const cv::Mat &float_image, const cv::Size_<int> &zoom_size, std::vector<cv::Mat> &cubic_planes);
 	eWaifu2xError LoadParameter(boost::shared_ptr<caffe::Net<float>> net, const std::string &param_path);
 	eWaifu2xError ConstractNet(boost::shared_ptr<caffe::Net<float>> &net, const std::string &model_path, const std::string &process);
-	eWaifu2xError ReconstructImage(boost::shared_ptr<caffe::Net<float>> net, cv::Mat &im);
+	eWaifu2xError ReconstructImage(std::vector<boost::shared_ptr<caffe::Net<float>>> nets, cv::Mat &im);
 
 public:
 	Waifu2x();
diff --git a/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj b/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj
index 01abfa8..6bf6272 100644
--- a/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj
+++ b/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj
@@ -55,7 +55,7 @@
       </PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>Disabled</Optimization>
-      <PreprocessorDefinitions>WIN32;_DEBUG;_WINDOWS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_WIN32_WINNT=0x0600;_DEBUG;_WINDOWS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Windows</SubSystem>
@@ -74,7 +74,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>WIN32;NDEBUG;_WINDOWS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_WIN32_WINNT=0x0600;NDEBUG;_WINDOWS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Windows</SubSystem>
@@ -95,6 +95,7 @@
     <ClCompile Include="Source.cpp" />
   </ItemGroup>
   <ItemGroup>
+    <ClInclude Include="..\common\tinypl.hpp" />
     <ClInclude Include="..\common\waifu2x.h" />
     <ClInclude Include="CControl.h" />
     <ClInclude Include="CDialog.h" />
diff --git a/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj.filters b/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj.filters
index 4282ea2..dbb02c5 100644
--- a/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj.filters
+++ b/waifu2x-caffe-gui/waifu2x-caffe-gui.vcxproj.filters
@@ -56,6 +56,9 @@
     <ClInclude Include="resource.h">
       <Filter>繝倥ャ繝�繝ｼ 繝輔ぃ繧､繝ｫ</Filter>
     </ClInclude>
+    <ClInclude Include="..\common\tinypl.hpp">
+      <Filter>繝倥ャ繝�繝ｼ 繝輔ぃ繧､繝ｫ</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ResourceCompile Include="Resource.rc">
diff --git a/waifu2x-caffe/waifu2x-caffe.vcxproj b/waifu2x-caffe/waifu2x-caffe.vcxproj
index b374766..b7295f8 100644
--- a/waifu2x-caffe/waifu2x-caffe.vcxproj
+++ b/waifu2x-caffe/waifu2x-caffe.vcxproj
@@ -55,7 +55,7 @@
       </PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>Disabled</Optimization>
-      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_WIN32_WINNT=0x0600;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -71,7 +71,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_WIN32_WINNT=0x0600;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -86,6 +86,7 @@
     <ClCompile Include="Source.cpp" />
   </ItemGroup>
   <ItemGroup>
+    <ClInclude Include="..\common\tinypl.hpp" />
     <ClInclude Include="..\common\waifu2x.h" />
   </ItemGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
diff --git a/waifu2x-caffe/waifu2x-caffe.vcxproj.filters b/waifu2x-caffe/waifu2x-caffe.vcxproj.filters
index 86e4980..4305a80 100644
--- a/waifu2x-caffe/waifu2x-caffe.vcxproj.filters
+++ b/waifu2x-caffe/waifu2x-caffe.vcxproj.filters
@@ -26,5 +26,8 @@
     <ClInclude Include="..\common\waifu2x.h">
       <Filter>繝倥ャ繝�繝ｼ 繝輔ぃ繧､繝ｫ</Filter>
     </ClInclude>
+    <ClInclude Include="..\common\tinypl.hpp">
+      <Filter>繝倥ャ繝�繝ｼ 繝輔ぃ繧､繝ｫ</Filter>
+    </ClInclude>
   </ItemGroup>
 </Project>
\ No newline at end of file