Not having read the C++ synchronisation primitives before, this async task queue is probably incorrect. But nevertheless, let's have a quick go at employing it — in a hideously thread unsafe fashion — for audio generation. What can possibly go wrong?

Concurrency/AsyncTaskQueue.cpp

@@ -0,0 +1,62 @@
+//
+//  AsyncTaskQueue.cpp
+//  Clock Signal
+//
+//  Created by Thomas Harte on 07/10/2016.
+//  Copyright © 2016 Thomas Harte. All rights reserved.
+//
+
+#include "AsyncTaskQueue.hpp"
+
+using namespace Concurrency;
+
+AsyncTaskQueue::AsyncTaskQueue() : should_destruct_(false)
+{
+	thread_.reset(new std::thread([this]() {
+		while(!should_destruct_)
+		{
+			std::function<void(void)> next_function;
+
+			queue_mutex_.lock();
+			if(!pending_tasks_.empty())
+			{
+				next_function = pending_tasks_.front();
+				pending_tasks_.pop_front();
+			}
+			queue_mutex_.unlock();
+
+			if(next_function)
+			{
+				next_function();
+			}
+			else
+			{
+				std::unique_lock<std::mutex> lock(queue_mutex_);
+				processing_condition_.wait(lock);
+				lock.unlock();
+			}
+		}
+	}));
+}
+
+AsyncTaskQueue::~AsyncTaskQueue()
+{
+	should_destruct_ = true;
+	enqueue([](){});
+}
+
+void AsyncTaskQueue::enqueue(std::function<void(void)> function)
+{
+	queue_mutex_.lock();
+	pending_tasks_.push_back(function);
+	queue_mutex_.unlock();
+
+	std::lock_guard<std::mutex> lock(queue_mutex_);
+	processing_condition_.notify_all();
+}
+
+void AsyncTaskQueue::synchronise()
+{
+	// TODO
+//	std::mutex
+}

Concurrency/AsyncTaskQueue.hpp

@@ -0,0 +1,39 @@
+//
+//  AsyncTaskQueue.hpp
+//  Clock Signal
+//
+//  Created by Thomas Harte on 07/10/2016.
+//  Copyright © 2016 Thomas Harte. All rights reserved.
+//
+
+#ifndef AsyncTaskQueue_hpp
+#define AsyncTaskQueue_hpp
+
+#include <memory>
+#include <thread>
+#include <list>
+#include <condition_variable>
+
+namespace Concurrency {
+
+class AsyncTaskQueue {
+
+	public:
+		AsyncTaskQueue();
+		~AsyncTaskQueue();
+
+		void enqueue(std::function<void(void)> function);
+		void synchronise();
+
+	private:
+		std::unique_ptr<std::thread> thread_;
+
+		std::mutex queue_mutex_;
+		std::list<std::function<void(void)>> pending_tasks_;
+		std::condition_variable processing_condition_;
+		bool should_destruct_;
+};
+
+}
+
+#endif /* Concurrency_hpp */

OSBindings/Mac/Clock Signal.xcodeproj/project.pbxproj

@@ -30,6 +30,7 @@
 		4B30512D1D989E2200B4FED8 /* Drive.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 4B30512B1D989E2200B4FED8 /* Drive.cpp */; };
 		4B3051301D98ACC600B4FED8 /* Plus3.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 4B30512E1D98ACC600B4FED8 /* Plus3.cpp */; };
 		4B37EE821D7345A6006A09A4 /* BinaryDump.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 4B37EE801D7345A6006A09A4 /* BinaryDump.cpp */; };
+		4B3940E71DA83C8300427841 /* AsyncTaskQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 4B3940E51DA83C8300427841 /* AsyncTaskQueue.cpp */; };
 		4B3BA0C31D318AEC005DD7A7 /* C1540Tests.swift in Sources */ = {isa = PBXBuildFile; fileRef = 4B3BA0C21D318AEB005DD7A7 /* C1540Tests.swift */; };
 		4B3BA0CE1D318B44005DD7A7 /* C1540Bridge.mm in Sources */ = {isa = PBXBuildFile; fileRef = 4B3BA0C61D318B44005DD7A7 /* C1540Bridge.mm */; };
 		4B3BA0CF1D318B44005DD7A7 /* MOS6522Bridge.mm in Sources */ = {isa = PBXBuildFile; fileRef = 4B3BA0C91D318B44005DD7A7 /* MOS6522Bridge.mm */; };
@@ -438,6 +439,8 @@
 		4B30512F1D98ACC600B4FED8 /* Plus3.hpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.h; name = Plus3.hpp; path = Electron/Plus3.hpp; sourceTree = "<group>"; };
 		4B37EE801D7345A6006A09A4 /* BinaryDump.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = BinaryDump.cpp; sourceTree = "<group>"; };
 		4B37EE811D7345A6006A09A4 /* BinaryDump.hpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.h; path = BinaryDump.hpp; sourceTree = "<group>"; };
+		4B3940E51DA83C8300427841 /* AsyncTaskQueue.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = AsyncTaskQueue.cpp; path = ../../Concurrency/AsyncTaskQueue.cpp; sourceTree = "<group>"; };
+		4B3940E61DA83C8300427841 /* AsyncTaskQueue.hpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.h; name = AsyncTaskQueue.hpp; path = ../../Concurrency/AsyncTaskQueue.hpp; sourceTree = "<group>"; };
 		4B3BA0C21D318AEB005DD7A7 /* C1540Tests.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = C1540Tests.swift; sourceTree = "<group>"; };
 		4B3BA0C51D318B44005DD7A7 /* C1540Bridge.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = C1540Bridge.h; sourceTree = "<group>"; };
 		4B3BA0C61D318B44005DD7A7 /* C1540Bridge.mm */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.objcpp; path = C1540Bridge.mm; sourceTree = "<group>"; };
@@ -1003,6 +1006,15 @@
 			name = Outputs;
 			sourceTree = "<group>";
 		};
+		4B3940E81DA83C8700427841 /* Concurrency */ = {
+			isa = PBXGroup;
+			children = (
+				4B3940E51DA83C8300427841 /* AsyncTaskQueue.cpp */,
+				4B3940E61DA83C8300427841 /* AsyncTaskQueue.hpp */,
+			);
+			name = Concurrency;
+			sourceTree = "<group>";
+		};
 		4B3BA0C41D318B44005DD7A7 /* Bridges */ = {
 			isa = PBXGroup;
 			children = (
@@ -1469,6 +1481,7 @@
 		4BB73E951B587A5100552FC2 = {
 			isa = PBXGroup;
 			children = (
+				4B3940E81DA83C8700427841 /* Concurrency */,
 				4BC76E6A1C98F43700E6EF73 /* Accelerate.framework */,
 				4BB73EA01B587A5100552FC2 /* Clock Signal */,
 				4BB73EB51B587A5100552FC2 /* Clock SignalTests */,
@@ -2185,6 +2198,7 @@
 				4B1E85751D170228001EF87D /* Typer.cpp in Sources */,
 				4BF829631D8F536B001BAE39 /* SSD.cpp in Sources */,
 				4B2E2D9D1C3A070400138695 /* Electron.cpp in Sources */,
+				4B3940E71DA83C8300427841 /* AsyncTaskQueue.cpp in Sources */,
 				4BAB62B81D3302CA00DF5BA0 /* PCMTrack.cpp in Sources */,
 				4B69FB3D1C4D908A00B5F0AA /* Tape.cpp in Sources */,
 				4B8FE2291DA1EDDF0090D3CE /* ElectronOptionsPanel.swift in Sources */,

Outputs/Speaker.hpp

@@ -14,6 +14,7 @@
 #include <time.h>
 #include "../SignalProcessing/Stepper.hpp"
 #include "../SignalProcessing/FIRFilter.hpp"
+#include "../Concurrency/AsyncTaskQueue.hpp"
 
 namespace Outputs {
 
@@ -25,7 +26,7 @@ namespace Outputs {
 	Intended to be a parent class, allowing descendants to pick the strategy by which input samples are mapped to
 	output samples.
 */
-class Speaker {
+class Speaker: public Concurrency::AsyncTaskQueue {
 	public:
 		class Delegate {
 			public:
@@ -124,49 +125,20 @@ template <class T> class Filter: public Speaker {
 	public:
 		void run_for_cycles(unsigned int input_cycles)
 		{
-			if(_coefficients_are_dirty) update_filter_coefficients();
+			enqueue([=]() {
+				unsigned int cycles_remaining = input_cycles;
+				if(_coefficients_are_dirty) update_filter_coefficients();
 
-			// if input and output rates exactly match, just accumulate results and pass on
-			if(_input_cycles_per_second == _output_cycles_per_second && _high_frequency_cut_off < 0.0)
-			{
-				while(input_cycles)
+				// if input and output rates exactly match, just accumulate results and pass on
+				if(_input_cycles_per_second == _output_cycles_per_second && _high_frequency_cut_off < 0.0)
 				{
-					unsigned int cycles_to_read = (unsigned int)(_buffer_size - _buffer_in_progress_pointer);
-					if(cycles_to_read > input_cycles) cycles_to_read = input_cycles;
-
-					static_cast<T *>(this)->get_samples(cycles_to_read, &_buffer_in_progress.get()[_buffer_in_progress_pointer]);
-					_buffer_in_progress_pointer += cycles_to_read;
-
-					// announce to delegate if full
-					if(_buffer_in_progress_pointer == _buffer_size)
+					while(cycles_remaining)
 					{
-						_buffer_in_progress_pointer = 0;
-						if(_delegate)
-						{
-							_delegate->speaker_did_complete_samples(this, _buffer_in_progress.get(), _buffer_size);
-						}
-					}
+						unsigned int cycles_to_read = (unsigned int)(_buffer_size - _buffer_in_progress_pointer);
+						if(cycles_to_read > cycles_remaining) cycles_to_read = cycles_remaining;
 
-					input_cycles -= cycles_to_read;
-				}
-
-				return;
-			}
-
-			// if the output rate is less than the input rate, use the filter
-			if(_input_cycles_per_second > _output_cycles_per_second)
-			{
-				while(input_cycles)
-				{
-					unsigned int cycles_to_read = (unsigned int)std::min((int)input_cycles, _number_of_taps - _input_buffer_depth);
-					static_cast<T *>(this)->get_samples(cycles_to_read, &_input_buffer.get()[_input_buffer_depth]);
-					input_cycles -= cycles_to_read;
-					_input_buffer_depth += cycles_to_read;
-
-					if(_input_buffer_depth == _number_of_taps)
-					{
-						_buffer_in_progress.get()[_buffer_in_progress_pointer] = _filter->apply(_input_buffer.get());
-						_buffer_in_progress_pointer++;
+						static_cast<T *>(this)->get_samples(cycles_to_read, &_buffer_in_progress.get()[_buffer_in_progress_pointer]);
+						_buffer_in_progress_pointer += cycles_to_read;
 
 						// announce to delegate if full
 						if(_buffer_in_progress_pointer == _buffer_size)
@@ -178,29 +150,61 @@ template <class T> class Filter: public Speaker {
 							}
 						}
 
-						// If the next loop around is going to reuse some of the samples just collected, use a memmove to
-						// preserve them in the correct locations (TODO: use a longer buffer to fix that) and don't skip
-						// anything. Otherwise skip as required to get to the next sample batch and don't expect to reuse.
-						uint64_t steps = _stepper->step();
-						if(steps < _number_of_taps)
-						{
-							int16_t *input_buffer = _input_buffer.get();
-							memmove(input_buffer, &input_buffer[steps], sizeof(int16_t) * ((size_t)_number_of_taps - (size_t)steps));
-							_input_buffer_depth -= steps;
-						}
-						else
-						{
-							if(steps > _number_of_taps)
-								static_cast<T *>(this)->skip_samples((unsigned int)steps - (unsigned int)_number_of_taps);
-							_input_buffer_depth = 0;
-						}
+						cycles_remaining -= cycles_to_read;
 					}
+
+					return;
 				}
 
-				return;
-			}
+				// if the output rate is less than the input rate, use the filter
+				if(_input_cycles_per_second > _output_cycles_per_second)
+				{
+					while(cycles_remaining)
+					{
+						unsigned int cycles_to_read = (unsigned int)std::min((int)cycles_remaining, _number_of_taps - _input_buffer_depth);
+						static_cast<T *>(this)->get_samples(cycles_to_read, &_input_buffer.get()[_input_buffer_depth]);
+						cycles_remaining -= cycles_to_read;
+						_input_buffer_depth += cycles_to_read;
+
+						if(_input_buffer_depth == _number_of_taps)
+						{
+							_buffer_in_progress.get()[_buffer_in_progress_pointer] = _filter->apply(_input_buffer.get());
+							_buffer_in_progress_pointer++;
+
+							// announce to delegate if full
+							if(_buffer_in_progress_pointer == _buffer_size)
+							{
+								_buffer_in_progress_pointer = 0;
+								if(_delegate)
+								{
+									_delegate->speaker_did_complete_samples(this, _buffer_in_progress.get(), _buffer_size);
+								}
+							}
+
+							// If the next loop around is going to reuse some of the samples just collected, use a memmove to
+							// preserve them in the correct locations (TODO: use a longer buffer to fix that) and don't skip
+							// anything. Otherwise skip as required to get to the next sample batch and don't expect to reuse.
+							uint64_t steps = _stepper->step();
+							if(steps < _number_of_taps)
+							{
+								int16_t *input_buffer = _input_buffer.get();
+								memmove(input_buffer, &input_buffer[steps], sizeof(int16_t) * ((size_t)_number_of_taps - (size_t)steps));
+								_input_buffer_depth -= steps;
+							}
+							else
+							{
+								if(steps > _number_of_taps)
+									static_cast<T *>(this)->skip_samples((unsigned int)steps - (unsigned int)_number_of_taps);
+								_input_buffer_depth = 0;
+							}
+						}
+					}
+
+					return;
+				}
 
 			// TODO: input rate is less than output rate
+			});
 		}
 
 	private: