diff --git a/src/application.cpp b/src/application.cpp
index 1d633a5..03003e3 100644
--- a/src/application.cpp
+++ b/src/application.cpp
@@ -47,6 +47,7 @@ static void StepControl() {
 				elements::TextButton(CLAY_STRING("Step"), style::actionButton, &StepSimulationButton);
 			}
 		}
+		elements::Toggle(CLAY_STRING("Threaded"), style::actionButton, simulation::threadedDesired);
 		elements::Toggle(CLAY_STRING("Show Changes"), style::actionButton, simulation::drawDebugInfo);
 		elements::Toggle(CLAY_STRING("Lock Speed"), style::actionButton, lockFramerate);
 	}
diff --git a/src/simulation.cpp b/src/simulation.cpp
index c068106..6c43b7b 100644
--- a/src/simulation.cpp
+++ b/src/simulation.cpp
@@ -28,6 +28,9 @@ struct ThreadWorkload {
 };
 
 bool drawDebugInfo{ true }; // draw next step's changes over current state
+bool threadedDesired{ true }; // multithread generation steps
+static bool actuallyThreaded{ false }; // notes whether or not currently ongoing tasks are multithreaded
+size_t threadsPerTask{ std::thread::hardware_concurrency() };
 
 static std::set<Cell> living{}; // The set of currently alive cells
 
@@ -87,11 +90,15 @@ static size_t CountNeighbors(Cell const &cell) {
 	return count;
 }
 
+// Block the thread until all tasks are done
+// after this function, all code in this file is synchronous, until another thread is started
 static void BlockUntilTasksDone() {
-	std::unique_lock lock{ tasksMutex };
-	while (tasks > 0) {
-		tasksChanged.wait(lock);
-	}
+	if (actuallyThreaded) {
+		std::unique_lock lock{ tasksMutex };
+		while (tasks > 0) {
+			tasksChanged.wait(lock);
+		}
+	} else while (tasks > 0) { }
 }
 
 static void TaskBegin() {
@@ -100,11 +107,16 @@ static void TaskBegin() {
 }
 
 static void TaskComplete() {
-	std::scoped_lock lock{ tasksMutex };
-	if (--tasks == 0) {
-		SDL_Log("Generation Complete %lfs", (double)(SDL_GetTicksNS() - generationStartTime) * 0.000000001);
+	if (actuallyThreaded) {
+		std::scoped_lock lock{ tasksMutex };
+		if (--tasks == 0) {
+			SDL_Log("Generation Complete %lfs", (double)(SDL_GetTicksNS() - generationStartTime) * 0.000000001);
+		}
+		tasksChanged.notify_all();
+	} else if (tasks == 3) {
+		SDL_Log("Generation complete %lfs", (double)(SDL_GetTicksNS() - generationStartTime) * 0.000000001);
+		tasks = 0;
 	}
-	tasksChanged.notify_all();
 }
 
 static void FindOverpopulated(std::shared_ptr<ThreadWorkload> wl) {
@@ -162,23 +174,11 @@ static void FindBorn(std::shared_ptr<ThreadWorkload> wl) {
 	TaskComplete();
 }
 
-static void PopulateChanges() {
-	static bool first_run{ true };
-#if SIM_MULTITHREADING
-	BlockUntilTasksDone();
-	// for some reason three tasks per thread is faster than one task per thread, i don't get it either
-	size_t const split{ std::thread::hardware_concurrency() };
-	if (first_run) {
-		first_run = false;
-		born.resize(split);
-		underpopulated.resize(split);
-		overpopulated.resize(split);
-	}
-
+static inline void TickGenerationThreaded() {
 	SDL_Log("Multithreading ON");
 	generationStartTime = SDL_GetTicksNS();
-	size_t const seg_length{ living.size() / split };
-	for (size_t i{ 0 }; i < split; ++i) {
+	size_t const seg_length{ living.size() / threadsPerTask };
+	for (size_t i{ 0 }; i < threadsPerTask; ++i) {
 		if (overpopulated[i] == nullptr)
 			overpopulated[i] = std::make_shared<ThreadWorkload>();
 		{ std::scoped_lock lock{ overpopulated[i]->mtx };
@@ -198,27 +198,42 @@ static void PopulateChanges() {
 			threading::tasks.ScheduleTask(std::bind(FindBorn, born[i]));
 		}
 	}
-#else
-	if (first_run) {
-		first_run = false;
-		overpopulated.emplace_back(std::make_shared<ThreadWorkload>());
-		overpopulated[0]->seg_idx = 0;
-		overpopulated[0]->seg_len = living.size();
-		underpopulated.emplace_back(std::make_shared<ThreadWorkload>());
-		underpopulated[0]->seg_idx = 0;
-		underpopulated[0]->seg_len = living.size();
-		born.emplace_back(std::make_shared<ThreadWorkload>());
-		born[0]->seg_idx = 0;
-		born[0]->seg_len = living.size();
-	}
+}
+
+static inline void TickGenerationSynchronous() {
 	SDL_Log("Multithreading OFF");
-	{ std::scoped_lock lock{ tasksMutex };
-		tasks = 3; }
 	generationStartTime = SDL_GetTicksNS();
+	born[0]->seg_idx = underpopulated[0]->seg_idx = overpopulated[0]->seg_idx = 0;
+	born[0]->seg_len = underpopulated[0]->seg_len = overpopulated[0]->seg_len = living.size();
+
 	FindOverpopulated(overpopulated[0]);
 	FindUnderpopulated(underpopulated[0]);
 	FindBorn(born[0]);
-#endif
+}
+
+static void PopulateChanges() {
+	static bool first_run{ true };
+	BlockUntilTasksDone();
+	// for some reason three tasks per thread is faster than one task per thread, i don't get it either
+	if (threadedDesired) {
+		if (first_run || !actuallyThreaded) {
+			actuallyThreaded = true;
+			first_run = false;
+			born.resize(threadsPerTask);
+			underpopulated.resize(threadsPerTask);
+			overpopulated.resize(threadsPerTask);
+		}
+		TickGenerationThreaded();
+	} else {
+		if (first_run || actuallyThreaded) {
+			actuallyThreaded = false;
+			first_run = false;
+			born.resize(1);
+			underpopulated.resize(1);
+			overpopulated.resize(1);
+		}
+		TickGenerationSynchronous();
+	}
 }
 
 void InitializeRandom(size_t livingChance, int64_t fillArea) {
diff --git a/src/simulation.h b/src/simulation.h
index df68d78..0dbbe50 100644
--- a/src/simulation.h
+++ b/src/simulation.h
@@ -34,6 +34,8 @@ public: CellRange(Cell begin, Cell end) : beginCell{ begin }, endCell{ end } {}
 };
 
 extern bool drawDebugInfo;
+extern bool threadedDesired;
+extern size_t threadsPerTask;
 
 void InitializeRandom(size_t livingChance, int64_t fillArea);
 void Step();