diff --git a/backend/BackendConnector.cpp b/backend/BackendConnector.cpp
new file mode 100644
index 0000000..f7a63fe
--- /dev/null
+++ b/backend/BackendConnector.cpp
@@ -0,0 +1,90 @@
+/*
+  * This program is free software: you can redistribute it and/or modify
+  * it under the terms of the GNU General Public License as published by
+  * the Free Software Foundation, either version 3 of the License, or
+  * any later version.
+  *
+  * This program is distributed in the hope that it will be useful,
+  * but WITHOUT ANY WARRANTY; without even the implied warranty of
+  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+  * GNU General Public License for more details.
+  *
+  * You should have received a copy of the GNU General Public License
+  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+  *
+  * Additional permission under GNU GPL version 3 section 7
+  *
+  * If you modify this Program, or any covered work, by linking or combining
+  * it with OpenSSL (or a modified version of that library), containing parts
+  * covered by the terms of OpenSSL License and SSLeay License, the licensors
+  * of this Program grant you additional permission to convey the resulting work.
+  *
+  */
+
+#include <assert.h>
+#include <cmath>
+#include <chrono>
+#include <cstring>
+#include <thread>
+#include <bitset>
+
+#include "IBackend.hpp"
+#include "BackendConnector.hpp"
+
+#include "cpu/minethd.h"
+#ifndef CONF_NO_CUDA
+#	include "nvidia/minethd.h"
+#endif
+#ifndef CONF_NO_OPENCL
+#	include "amd/minethd.h"
+#endif
+#include "miner_work.h"
+#include "GlobalStates.hpp"
+ #include <cstdlib>
+
+#include "Plugin.hpp"
+
+namespace xmrstak
+{
+
+
+bool BackendConnector::self_test()
+{
+	
+	return true;
+}
+
+std::vector<IBackend*>* BackendConnector::thread_starter(miner_work& pWork)
+{
+    GlobalStates::iGlobalJobNo = 0;
+	GlobalStates::iConsumeCnt = 0;
+
+
+	std::vector<IBackend*>* pvThreads = new std::vector<IBackend*>;
+
+	auto cpuThreads = cpu::minethd::thread_starter(pvThreads->size(), pWork);
+	pvThreads->insert(std::end(*pvThreads), std::begin(cpuThreads), std::end(cpuThreads));
+	if(cpuThreads.size() == 0)
+		printer::inst()->print_msg(L0, "WARNING: backend CPU disabled.");
+
+#ifndef CONF_NO_CUDA
+	Plugin nvidiaPlugin("NVIDIA", "./libxmrstak_cuda_backend");
+	std::vector<IBackend*>* nvidiaThreads = nvidiaPlugin.startBackend(pvThreads->size(), pWork);
+	pvThreads->insert(std::end(*pvThreads), std::begin(*nvidiaThreads), std::end(*nvidiaThreads));
+	if(nvidiaThreads->size() == 0)
+		printer::inst()->print_msg(L0, "WARNING: backend NVIDIA disabled.");
+#endif
+
+#ifndef CONF_NO_OPENCL
+	Plugin amdPlugin("AMD", "./libxmrstak_opencl_backend");
+	std::vector<IBackend*>* amdThreads = amdPlugin.startBackend(pvThreads->size(), pWork);
+	pvThreads->insert(std::end(*pvThreads), std::begin(*amdThreads), std::end(*amdThreads));
+	if(amdThreads->size() == 0)
+		printer::inst()->print_msg(L0, "WARNING: backend AMD disabled.");
+#endif
+	
+	GlobalStates::iThreadCount = pvThreads->size();
+	return pvThreads;
+}
+
+} // namepsace xmrstak
diff --git a/backend/BackendConnector.hpp b/backend/BackendConnector.hpp
new file mode 100644
index 0000000..024d9b1
--- /dev/null
+++ b/backend/BackendConnector.hpp
@@ -0,0 +1,18 @@
+#pragma once
+#include <thread>
+#include <vector>
+#include <atomic>
+#include <mutex>
+#include "IBackend.hpp"
+#include "miner_work.h"
+
+namespace xmrstak
+{
+
+	struct BackendConnector
+	{
+		static std::vector<IBackend*>* thread_starter(miner_work& pWork);
+		static bool self_test();
+	};
+
+} // namepsace xmrstak
diff --git a/backend/GlobalStates.cpp b/backend/GlobalStates.cpp
new file mode 100644
index 0000000..99356b5
--- /dev/null
+++ b/backend/GlobalStates.cpp
@@ -0,0 +1,55 @@
+/*
+  * This program is free software: you can redistribute it and/or modify
+  * it under the terms of the GNU General Public License as published by
+  * the Free Software Foundation, either version 3 of the License, or
+  * any later version.
+  *
+  * This program is distributed in the hope that it will be useful,
+  * but WITHOUT ANY WARRANTY; without even the implied warranty of
+  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+  * GNU General Public License for more details.
+  *
+  * You should have received a copy of the GNU General Public License
+  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+  *
+  * Additional permission under GNU GPL version 3 section 7
+  *
+  * If you modify this Program, or any covered work, by linking or combining
+  * it with OpenSSL (or a modified version of that library), containing parts
+  * covered by the terms of OpenSSL License and SSLeay License, the licensors
+  * of this Program grant you additional permission to convey the resulting work.
+  *
+  */
+
+#include <assert.h>
+#include <cmath>
+#include <chrono>
+#include <cstring>
+
+
+#include "miner_work.h"
+#include "GlobalStates.hpp"
+
+namespace xmrstak
+{
+
+std::atomic<uint64_t> GlobalStates::iGlobalJobNo;
+std::atomic<uint64_t> GlobalStates::iConsumeCnt; //Threads get jobs as they are initialized
+miner_work GlobalStates::oGlobalWork;
+uint64_t GlobalStates::iThreadCount = 0;
+
+void GlobalStates::switch_work(miner_work& pWork)
+{
+	// iConsumeCnt is a basic lock-like polling mechanism just in case we happen to push work
+	// faster than threads can consume them. This should never happen in real life.
+	// Pool cant physically send jobs faster than every 250ms or so due to net latency.
+
+	while (iConsumeCnt.load(std::memory_order_seq_cst) < iThreadCount)
+		std::this_thread::sleep_for(std::chrono::milliseconds(100));
+
+	oGlobalWork = pWork;
+	iConsumeCnt.store(0, std::memory_order_seq_cst);
+	iGlobalJobNo++;
+}
+
+} // namepsace xmrstak
diff --git a/backend/GlobalStates.hpp b/backend/GlobalStates.hpp
new file mode 100644
index 0000000..293009d
--- /dev/null
+++ b/backend/GlobalStates.hpp
@@ -0,0 +1,20 @@
+#pragma once
+#include <atomic>
+#include "miner_work.h"
+
+namespace xmrstak
+{
+
+struct GlobalStates
+{
+
+	static void switch_work(miner_work& pWork);
+
+	static miner_work oGlobalWork;
+	static std::atomic<uint64_t> iGlobalJobNo;
+	static std::atomic<uint64_t> iConsumeCnt;
+	static uint64_t iThreadCount;
+	
+};
+
+} // namepsace xmrstak
diff --git a/backend/IBackend.hpp b/backend/IBackend.hpp
new file mode 100644
index 0000000..5037028
--- /dev/null
+++ b/backend/IBackend.hpp
@@ -0,0 +1,20 @@
+#pragma once
+
+#include <atomic>
+#include <cstdint>
+
+
+namespace xmrstak
+{
+
+	struct IBackend
+	{
+		std::atomic<uint64_t> iHashCount;
+		std::atomic<uint64_t> iTimestamp;
+
+		IBackend() : iHashCount(0), iTimestamp(0)
+		{
+		}
+	};
+
+} // namepsace xmrstak
diff --git a/backend/Plugin.hpp b/backend/Plugin.hpp
new file mode 100644
index 0000000..762f955
--- /dev/null
+++ b/backend/Plugin.hpp
@@ -0,0 +1,90 @@
+#pragma once
+#include <thread>
+#include <atomic>
+#include <vector>
+#include <string>
+#include "IBackend.hpp"
+#include <iostream>
+
+#ifndef USE_PRECOMPILED_HEADERS
+#ifdef WIN32
+#include <direct.h>
+#include <windows.h>
+#else
+#include <sys/types.h>
+#include <dlfcn.h>
+#endif
+#include <iostream>
+#endif
+
+namespace xmrstak
+{
+
+struct Plugin
+{
+
+	Plugin(const std::string backendName, const std::string libName) : fn_starterBackend(nullptr), m_backendName(backendName)
+	{
+#ifdef WIN32
+		libBackend = LoadLibrary(TEXT((libName + ".dll").c_str()));
+#else
+		libBackend = dlopen((libName + ".so").c_str(), RTLD_LAZY);
+#endif
+		if (!libBackend)
+		{
+			std::cerr << "WARNING: "<< m_backendName <<" cannot load backend library: " << dlerror() << std::endl;
+			return;
+		}
+
+#ifdef WIN32
+		fn_starterBackend = (func_t*) GetProcAddress(libBackend, "xmrstak_start_backend");
+		if (!fn_starterBackend)
+		{
+			std::cerr << "WARNING: backend plugin " << libName << " contains no entry 'xmrstak_start_backend'" << std::endl;
+		}
+#else
+		// reset last error
+		dlerror();
+		fn_starterBackend = (starterBackend_t) dlsym(libBackend, "xmrstak_start_backend");
+		const char* dlsym_error = dlerror();
+		if (dlsym_error)
+		{
+			std::cerr << "WARNING: backend plugin " << libName << " contains no entry 'xmrstak_start_backend'" << std::endl;
+		}
+#endif
+	}
+
+	std::vector<IBackend*>* startBackend(uint32_t threadOffset, miner_work& pWork)
+	{
+		if(fn_starterBackend == nullptr)
+		{
+			std::vector<IBackend*>* pvThreads = new std::vector<IBackend*>();
+			std::cerr << "WARNING: " << m_backendName << " Backend disabled"<< std::endl;
+			return pvThreads;
+		}
+
+		return fn_starterBackend(threadOffset, pWork);
+	}
+
+	std::string m_backendName;
+
+	typedef std::vector<IBackend*>* (*starterBackend_t)(uint32_t threadOffset, miner_work& pWork);
+
+	starterBackend_t fn_starterBackend;
+
+#ifdef WIN32
+    HINSTANCE libBackend;
+#else
+    void *libBackend;
+#endif
+
+/* \todo add unload to destructor and change usage of Plugin that libs keeped open until the miner endss
+#ifdef WIN32
+    FreeLibrary(libBackend);
+#else
+    dlclose(libBackend);
+#endif
+ * */
+};
+
+} // namepsace xmrstak
diff --git a/backend/miner_work.h b/backend/miner_work.h
new file mode 100644
index 0000000..b922f4a
--- /dev/null
+++ b/backend/miner_work.h
@@ -0,0 +1,102 @@
+#pragma once
+#include <thread>
+#include <atomic>
+#include <mutex>
+#include <cstdint>
+#include <climits>
+#include <iostream>
+
+namespace xmrstak
+{
+	// only allowed for unsigned value \todo add static assert
+	template<typename T>
+	T reverseBits(T value)
+	{
+		/* init with value (to get LSB) */
+		T result = value;
+		/* extra shift needed at end */
+		int s = sizeof(T) * CHAR_BIT - 1;
+		for (value >>= 1; value; value >>= 1)
+		{
+			result <<= 1;
+			result |= value & 1;
+			s--;
+		}
+		/* shift when values highest bits are zero */
+		result <<= s;
+		return result;
+	}
+
+	struct miner_work
+	{
+		char        sJobID[64];
+		uint8_t     bWorkBlob[112];
+		uint32_t    iWorkSize;
+		uint32_t    iResumeCnt;
+		uint64_t    iTarget;
+		// \todo remove workaround needed for amd
+		uint32_t    iTarget32;
+		bool        bNiceHash;
+		bool        bStall;
+		size_t      iPoolId;
+
+		miner_work() : iWorkSize(0), bNiceHash(false), bStall(true), iPoolId(0) { }
+
+		miner_work(const char* sJobID, const uint8_t* bWork, uint32_t iWorkSize, uint32_t iResumeCnt,
+			uint64_t iTarget, size_t iPoolId) : iWorkSize(iWorkSize), iResumeCnt(iResumeCnt),
+			iTarget(iTarget), bNiceHash(false), bStall(false), iPoolId(iPoolId)
+		{
+			assert(iWorkSize <= sizeof(bWorkBlob));
+			memcpy(this->sJobID, sJobID, sizeof(miner_work::sJobID));
+			memcpy(this->bWorkBlob, bWork, iWorkSize);
+		}
+
+		miner_work(miner_work const&) = delete;
+
+		miner_work& operator=(miner_work const& from)
+		{
+			assert(this != &from);
+
+			iWorkSize = from.iWorkSize;
+			iResumeCnt = from.iResumeCnt;
+			iTarget = from.iTarget;
+			iTarget32 = from.iTarget32;
+			bNiceHash = from.bNiceHash;
+			bStall = from.bStall;
+			iPoolId = from.iPoolId;
+
+			assert(iWorkSize <= sizeof(bWorkBlob));
+			memcpy(sJobID, from.sJobID, sizeof(sJobID));
+			memcpy(bWorkBlob, from.bWorkBlob, iWorkSize);
+
+			return *this;
+		}
+
+		miner_work(miner_work&& from) : iWorkSize(from.iWorkSize), iTarget(from.iTarget),iTarget32(from.iTarget32),
+			bStall(from.bStall), iPoolId(from.iPoolId)
+		{
+			assert(iWorkSize <= sizeof(bWorkBlob));
+			memcpy(sJobID, from.sJobID, sizeof(sJobID));
+			memcpy(bWorkBlob, from.bWorkBlob, iWorkSize);
+		}
+
+		miner_work& operator=(miner_work&& from)
+		{
+			assert(this != &from);
+
+			iWorkSize = from.iWorkSize;
+			iResumeCnt = from.iResumeCnt;
+			iTarget = from.iTarget;
+			iTarget32 = from.iTarget32;
+			bNiceHash = from.bNiceHash;
+			bStall = from.bStall;
+			iPoolId = from.iPoolId;
+
+			assert(iWorkSize <= sizeof(bWorkBlob));
+			memcpy(sJobID, from.sJobID, sizeof(sJobID));
+			memcpy(bWorkBlob, from.bWorkBlob, iWorkSize);
+
+			return *this;
+		}
+	};
+} // namepsace xmrstak