diff --git a/DDrawCompat/Config/Config.cpp b/DDrawCompat/Config/Config.cpp
index 6615246..beaf024 100644
--- a/DDrawCompat/Config/Config.cpp
+++ b/DDrawCompat/Config/Config.cpp
@@ -9,6 +9,7 @@ namespace Config
 	Settings::BltFilter bltFilter;
 	Settings::ConfigHotKey configHotKey;
 	Settings::CpuAffinity cpuAffinity;
+	Settings::CpuAffinityRotation cpuAffinityRotation;
 	Settings::DesktopColorDepth desktopColorDepth;
 	Settings::DisplayFilter displayFilter;
 	Settings::DisplayRefreshRate displayRefreshRate;
diff --git a/DDrawCompat/Config/Config.h b/DDrawCompat/Config/Config.h
index d571c23..d42cba6 100644
--- a/DDrawCompat/Config/Config.h
+++ b/DDrawCompat/Config/Config.h
@@ -7,6 +7,7 @@
 #include <Config/Settings/BltFilter.h>
 #include <Config/Settings/ConfigHotKey.h>
 #include <Config/Settings/CpuAffinity.h>
+#include <Config/Settings/CpuAffinityRotation.h>
 #include <Config/Settings/DesktopColorDepth.h>
 #include <Config/Settings/DisplayFilter.h>
 #include <Config/Settings/DisplayRefreshRate.h>
@@ -43,6 +44,7 @@ namespace Config
 	extern Settings::BltFilter bltFilter;
 	extern Settings::ConfigHotKey configHotKey;
 	extern Settings::CpuAffinity cpuAffinity;
+	extern Settings::CpuAffinityRotation cpuAffinityRotation;
 	extern Settings::DesktopColorDepth desktopColorDepth;
 	extern Settings::DisplayFilter displayFilter;
 	extern Settings::DisplayRefreshRate displayRefreshRate;
diff --git a/DDrawCompat/Config/Settings/CpuAffinityRotation.h b/DDrawCompat/Config/Settings/CpuAffinityRotation.h
new file mode 100644
index 0000000..8553b20
--- /dev/null
+++ b/DDrawCompat/Config/Settings/CpuAffinityRotation.h
@@ -0,0 +1,18 @@
+#pragma once
+
+#include <Config/EnumSetting.h>
+
+namespace Config
+{
+	namespace Settings
+	{
+		class CpuAffinityRotation : public MappedSetting<bool>
+		{
+		public:
+			CpuAffinityRotation()
+				: MappedSetting("CpuAffinityRotation", "on", { {"off", false}, {"on", true} })
+			{
+			}
+		};
+	}
+}
diff --git a/DDrawCompat/DDraw/RealPrimarySurface.cpp b/DDrawCompat/DDraw/RealPrimarySurface.cpp
index 2e9181c..5b3a1fe 100644
--- a/DDrawCompat/DDraw/RealPrimarySurface.cpp
+++ b/DDrawCompat/DDraw/RealPrimarySurface.cpp
@@ -33,6 +33,7 @@
 #include <Overlay/ConfigWindow.h>
 #include <Overlay/StatsWindow.h>
 #include <Win32/DisplayMode.h>
+#include <Win32/Thread.h>
 
 namespace
 {
@@ -418,6 +419,7 @@ namespace
 		int msUntilUpdateReady = 0;
 		while (true)
 		{
+			Win32::Thread::rotateCpuAffinity();
 			if (msUntilUpdateReady > 0)
 			{
 				Sleep(1);
diff --git a/DDrawCompat/DDrawCompat.vcxproj b/DDrawCompat/DDrawCompat.vcxproj
index ab2693c..9cf3467 100644
--- a/DDrawCompat/DDrawCompat.vcxproj
+++ b/DDrawCompat/DDrawCompat.vcxproj
@@ -166,6 +166,7 @@
     <ClInclude Include="Config\Settings\BltFilter.h" />
     <ClInclude Include="Config\Settings\ConfigHotKey.h" />
     <ClInclude Include="Config\Settings\CpuAffinity.h" />
+    <ClInclude Include="Config\Settings\CpuAffinityRotation.h" />
     <ClInclude Include="Config\Settings\DesktopColorDepth.h" />
     <ClInclude Include="Config\Settings\DisplayFilter.h" />
     <ClInclude Include="Config\Settings\DisplayRefreshRate.h" />
diff --git a/DDrawCompat/DDrawCompat.vcxproj.filters b/DDrawCompat/DDrawCompat.vcxproj.filters
index 9b080e1..6cc8a85 100644
--- a/DDrawCompat/DDrawCompat.vcxproj.filters
+++ b/DDrawCompat/DDrawCompat.vcxproj.filters
@@ -633,6 +633,9 @@
     <ClInclude Include="Config\Settings\TerminateHotKey.h">
       <Filter>Header Files\Config\Settings</Filter>
     </ClInclude>
+    <ClInclude Include="Config\Settings\CpuAffinityRotation.h">
+      <Filter>Header Files\Config\Settings</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="Gdi\Gdi.cpp">
diff --git a/DDrawCompat/Dll/Dll.cpp b/DDrawCompat/Dll/Dll.cpp
index 866ddaf..f4d96bd 100644
--- a/DDrawCompat/Dll/Dll.cpp
+++ b/DDrawCompat/Dll/Dll.cpp
@@ -1,5 +1,6 @@
 #include <process.h>
 
+#include <Common/Hook.h>
 #include <Dll/Dll.h>
 
 namespace Dll
@@ -16,6 +17,7 @@ namespace Dll
 		if (thread)
 		{
 			SetThreadPriority(thread, priority);
+			CALL_ORIG_FUNC(SetThreadPriorityBoost)(thread, FALSE);
 		}
 		return thread;
 	}
diff --git a/DDrawCompat/Win32/Thread.cpp b/DDrawCompat/Win32/Thread.cpp
index 38c6d7f..150c005 100644
--- a/DDrawCompat/Win32/Thread.cpp
+++ b/DDrawCompat/Win32/Thread.cpp
@@ -1,12 +1,181 @@
+#include <bitset>
+#include <map>
+#include <sstream>
+#include <vector>
+
 #include <Windows.h>
 
 #include <Common/Hook.h>
 #include <Common/Log.h>
+#include <Common/Time.h>
 #include <Config/Config.h>
+#include <Dll/Dll.h>
 #include <Win32/Thread.h>
 
 namespace
 {
+	DWORD g_cpuAffinity = 0;
+	bool g_cpuAffinityRotationEnabled = false;
+	std::map<BYTE, BYTE> g_nextProcessor;
+
+	std::string maskToString(ULONG_PTR mask);
+	void setNextProcessorSet(DWORD fromSet, DWORD toSet);
+
+	std::map<BYTE, std::vector<DWORD>> getProcessorSets()
+	{
+		std::map<BYTE, std::vector<DWORD>> result;
+		DWORD primaryProcessorGroup = 0;
+		auto getThreadIdealProcessorEx = reinterpret_cast<decltype(&GetThreadIdealProcessorEx)>(
+			Compat::getProcAddress(GetModuleHandle("kernel32"), "GetThreadIdealProcessorEx"));
+		if (getThreadIdealProcessorEx)
+		{
+			PROCESSOR_NUMBER pn = {};
+			getThreadIdealProcessorEx(GetCurrentThread(), &pn);
+			primaryProcessorGroup = pn.Group;
+			LOG_INFO << "Primary CPU group number: " << primaryProcessorGroup;
+		}
+
+		auto getLogicalProcessorInformationEx = reinterpret_cast<decltype(&GetLogicalProcessorInformationEx)>(
+			Compat::getProcAddress(GetModuleHandle("kernel32"), "GetLogicalProcessorInformationEx"));
+		if (getLogicalProcessorInformationEx)
+		{
+			DWORD size = 0;
+			std::vector<BYTE> buffer;
+			getLogicalProcessorInformationEx(RelationProcessorCore, nullptr, &size);
+			buffer.resize(size);
+			getLogicalProcessorInformationEx(RelationProcessorCore,
+				reinterpret_cast<SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*>(buffer.data()), &size);
+
+			DWORD offset = 0;
+			while (offset < size)
+			{
+				auto& pi = *reinterpret_cast<SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*>(buffer.data() + offset);
+				for (DWORD i = 0; i < pi.Processor.GroupCount; ++i)
+				{
+					if (primaryProcessorGroup == pi.Processor.GroupMask[i].Group)
+					{
+						result[pi.Processor.EfficiencyClass].push_back(pi.Processor.GroupMask[i].Mask);
+					}
+				}
+				offset += pi.Size;
+			}
+		}
+		else
+		{
+			DWORD size = 0;
+			GetLogicalProcessorInformation(nullptr, &size);
+			std::vector<SYSTEM_LOGICAL_PROCESSOR_INFORMATION> processorInfo;
+			processorInfo.resize(size / sizeof(processorInfo[0]));
+			GetLogicalProcessorInformation(processorInfo.data(), &size);
+
+			for (auto& pi : processorInfo)
+			{
+				if (RelationProcessorCore == pi.Relationship)
+				{
+					result[0].push_back(pi.ProcessorMask);
+				}
+			}
+		}
+
+		if (result.empty())
+		{
+			LOG_INFO << "ERROR: Failed to detect CPU topology!";
+			return result;
+		}
+
+		for (auto it = result.rbegin(); it != result.rend(); ++it)
+		{
+			LOG_INFO << "Physical to logical CPU core mapping for efficiency class " << static_cast<DWORD>(it->first) << ':';
+			DWORD core = 0;
+			for (auto& set : it->second)
+			{
+				++core;
+				LOG_INFO << "  Physical core #" << core << ": " << maskToString(set);
+			}
+		}
+		return result;
+	}
+
+	void initNextProcessorMap(const std::map<BYTE, std::vector<DWORD>>& processorSets)
+	{
+		for (auto& ps : processorSets)
+		{
+			for (BYTE fromIndex = 0; fromIndex < ps.second.size(); ++fromIndex)
+			{
+				auto bitCount = std::bitset<32>(ps.second[fromIndex]).count();
+				bool found = false;
+				for (BYTE toIndex = fromIndex + 1; toIndex < ps.second.size() && !found; ++toIndex)
+				{
+					if (bitCount == std::bitset<32>(ps.second[toIndex]).count())
+					{
+						setNextProcessorSet(ps.second[fromIndex], ps.second[toIndex]);
+						found = true;
+					}
+				}
+				for (BYTE toIndex = 0; toIndex < fromIndex && !found; ++toIndex)
+				{
+					if (bitCount == std::bitset<32>(ps.second[toIndex]).count())
+					{
+						setNextProcessorSet(ps.second[fromIndex], ps.second[toIndex]);
+						found = true;
+					}
+				}
+				if (!found)
+				{
+					setNextProcessorSet(ps.second[fromIndex], ps.second[fromIndex]);
+				}
+			}
+		}
+	}
+
+	std::string maskToString(ULONG_PTR mask)
+	{
+		std::ostringstream oss;
+		for (BYTE i = 0; i < 32; ++i)
+		{
+			if (mask & (1U << i))
+			{
+				oss << i + 1 << ", ";
+			}
+		}
+		return oss.str().substr(0, max(0, oss.str().length() - 2));
+	}
+
+	DWORD rotateMask(DWORD mask)
+	{
+		DWORD result = 0;
+		for (BYTE i = 0; i < 32; ++i)
+		{
+			if (mask & (1U << i))
+			{
+				auto it = g_nextProcessor.find(i);
+				result |= 1U << (it != g_nextProcessor.end() ? it->second : i);
+			}
+		}
+		return result;
+	}
+
+	void setNextProcessorSet(DWORD fromSet, DWORD toSet)
+	{
+		BYTE from = 0;
+		BYTE to = 0;
+		while (0 != fromSet)
+		{
+			while (!(fromSet & (1U << from)))
+			{
+				++from;
+			}
+			while (!(toSet & (1U << to)))
+			{
+				++to;
+			}
+			g_nextProcessor[from] = to;
+			fromSet &= ~(1U << from);
+			++from;
+			++to;
+		}
+	}
+
 	BOOL WINAPI setProcessAffinityMask(HANDLE hProcess, DWORD_PTR dwProcessAffinityMask)
 	{
 		LOG_FUNC("SetProcessAffinityMask", hProcess, Compat::hex(dwProcessAffinityMask));
@@ -44,22 +213,49 @@ namespace Win32
 	{
 		void applyConfig()
 		{
-			auto cpuAffinity = Config::cpuAffinity.get();
-			if (0 != cpuAffinity)
-			{
-				SYSTEM_INFO si = {};
-				GetSystemInfo(&si);
-				const unsigned cpuCount = min(si.dwNumberOfProcessors, 32);
-				cpuAffinity &= UINT_MAX >> (32 - cpuCount);
+			initNextProcessorMap(getProcessorSets());
 
-				if (0 == cpuAffinity || !CALL_ORIG_FUNC(SetProcessAffinityMask)(GetCurrentProcess(), cpuAffinity))
+			g_cpuAffinity = Config::cpuAffinity.get();
+			if (0 != g_cpuAffinity)
+			{
+				for (BYTE i = 0; i < 32; ++i)
 				{
-					LOG_INFO << (0 == cpuAffinity ? "Invalid" : "Failed to set") << " CPU affinity, falling back to default";
-					Config::cpuAffinity.reset();
-					CALL_ORIG_FUNC(SetProcessAffinityMask)(GetCurrentProcess(), Config::cpuAffinity.get());
+					if (g_nextProcessor.find(i) == g_nextProcessor.end())
+					{
+						g_cpuAffinity &= ~(1U << i);
+					}
+				}
+
+				if (0 == g_cpuAffinity)
+				{
+					LOG_INFO << "Warning: CPU affinity setting doesn't match any existing logical cores, using default";
+					g_cpuAffinity = 1;
 				}
 			}
 
+			DWORD processMask = 0;
+			DWORD systemMask = 0;
+			GetProcessAffinityMask(GetCurrentProcess(), &processMask, &systemMask);
+			LOG_INFO << "Current CPU affinity: " << maskToString(processMask);
+
+			if (0 == g_cpuAffinity && 0 != processMask && processMask != systemMask)
+			{
+				g_cpuAffinity = processMask;
+			}
+
+			if (0 != g_cpuAffinity)
+			{
+				LOG_INFO << "Applying configured CPU affinity: " << maskToString(g_cpuAffinity);
+				if (!CALL_ORIG_FUNC(SetProcessAffinityMask)(GetCurrentProcess(), g_cpuAffinity))
+				{
+					LOG_INFO << "ERROR: Failed to set CPU affinity";
+					g_cpuAffinity = 0;
+				}
+			}
+
+			g_cpuAffinityRotationEnabled = Config::cpuAffinityRotation.get() && rotateMask(g_cpuAffinity) != g_cpuAffinity;
+			LOG_INFO << "CPU affinity rotation is " << (g_cpuAffinityRotationEnabled ? "enabled" : "disabled");
+
 			switch (Config::threadPriorityBoost.get())
 			{
 			case Config::Settings::ThreadPriorityBoost::OFF:
@@ -83,5 +279,27 @@ namespace Win32
 			HOOK_FUNCTION(kernel32, SetProcessPriorityBoost, setProcessPriorityBoost);
 			HOOK_FUNCTION(kernel32, SetThreadPriorityBoost, setThreadPriorityBoost);
 		}
+
+		void rotateCpuAffinity()
+		{
+			if (!g_cpuAffinityRotationEnabled)
+			{
+				return;
+			}
+
+			static auto g_qpcLastRotation = Time::queryPerformanceCounter();
+			auto qpcNow = Time::queryPerformanceCounter();
+			if (qpcNow - g_qpcLastRotation < Time::g_qpcFrequency / 10)
+			{
+				return;
+			}
+			g_qpcLastRotation = qpcNow;
+
+			g_cpuAffinity = rotateMask(g_cpuAffinity);
+			if (!CALL_ORIG_FUNC(SetProcessAffinityMask)(GetCurrentProcess(), g_cpuAffinity))
+			{
+				LOG_ONCE("ERROR: Failed to set rotated CPU affinity: " << maskToString(g_cpuAffinity));
+			}
+		}
 	}
 }
diff --git a/DDrawCompat/Win32/Thread.h b/DDrawCompat/Win32/Thread.h
index e0ad174..1e00bba 100644
--- a/DDrawCompat/Win32/Thread.h
+++ b/DDrawCompat/Win32/Thread.h
@@ -6,5 +6,6 @@ namespace Win32
 	{
 		void applyConfig();
 		void installHooks();
+		void rotateCpuAffinity();
 	}
 }