(svn r17776) -Codechange: [SDL] make "update the video card"-process asynchronious. Profiling with gprof etc. hasn't shown us that DrawSurfaceToScreen takes a significant amount of CPU; only using TIC/TOC it became apparant that it was a heavy CPU-cycle user or that it was waiting for something.

The benefit of making this function asynchronious ranges from 2%-25% (real time) during fast forward on dual core/hyperthreading-enabled CPUs; 8bpp improvements are, in my test cases, significantly smaller than 32bpp improvements.
On single core non-hyperthreading-enabled CPUs the extra locking/scheduling costs up to 1% extra realtime in fast forward. You can use -v sdl:no_threads to disable threading and undo this loss.
During normal non-fast-forwarded games the benefit/costs are negligable except when the gameloop takes more than about 90% of the time of a tick.
Note that allegro's performance does not improve with this system, likely due to their way of getting data to the video card. It is not implemented for the OS X/Windows video backends, unless (ofcourse) SDL is used there.
Funny is that the performance of the 32bpp(-anim) blitter is, at least in some test cases, significantly faster (more than 10%) than the 8bpp(-optimized) blitter when looking at real time in fast forward on a dual core CPU; it was slower.
The idea comes from a paper/report by Idar Borlaug and Knut Imar Hagen.

src/thread/thread.h

@@ -68,6 +68,19 @@ public:
 	 * End of the critical section
 	 */
 	virtual void EndCritical() = 0;
+
+	/**
+	 * Wait for a signal to be send.
+	 * @pre You must be in the critical section.
+	 * @note While waiting the critical section is left.
+	 * @post You will be in the critical section.
+	 */
+	virtual void WaitForSignal() = 0;
+
+	/**
+	 * Send a signal and wake the 'thread' that was waiting for it.
+	 */
+	virtual void SendSignal() = 0;
 };
 
 #endif /* THREAD_H */

src/thread/thread_none.cpp

@@ -23,6 +23,8 @@ class ThreadMutex_None : public ThreadMutex {
 public:
 	virtual void BeginCritical() {}
 	virtual void EndCritical() {}
+	virtual void WaitForSignal() {}
+	virtual void SendSignal() {}
 };
 
 /* static */ ThreadMutex *ThreadMutex::New()

src/thread/thread_pthread.cpp

@@ -12,6 +12,7 @@
 #include "../stdafx.h"
 #include "thread.h"
 #include <pthread.h>
+#include <errno.h>
 
 /**
  * POSIX pthread version for ThreadObject.
@@ -95,16 +96,21 @@ private:
 class ThreadMutex_pthread : public ThreadMutex {
 private:
 	pthread_mutex_t mutex;
+	pthread_cond_t condition;
 
 public:
 	ThreadMutex_pthread()
 	{
 		pthread_mutex_init(&this->mutex, NULL);
+		pthread_cond_init(&this->condition, NULL);
 	}
 
 	/* virtual */ ~ThreadMutex_pthread()
 	{
-		pthread_mutex_destroy(&this->mutex);
+		int err = pthread_cond_destroy(&this->condition);
+		assert(err != EBUSY);
+		err = pthread_mutex_destroy(&this->mutex);
+		assert(err != EBUSY);
 	}
 
 	/* virtual */ void BeginCritical()
@@ -116,6 +122,16 @@ public:
 	{
 		pthread_mutex_unlock(&this->mutex);
 	}
+
+	/* virtual */ void WaitForSignal()
+	{
+		pthread_cond_wait(&this->condition, &this->mutex);
+	}
+
+	/* virtual */ void SendSignal()
+	{
+		pthread_cond_signal(&this->condition);
+	}
 };
 
 /* static */ ThreadMutex *ThreadMutex::New()

src/thread/thread_win32.cpp

@@ -107,16 +107,19 @@ private:
 class ThreadMutex_Win32 : public ThreadMutex {
 private:
 	CRITICAL_SECTION critical_section;
+	HANDLE event;
 
 public:
 	ThreadMutex_Win32()
 	{
 		InitializeCriticalSection(&this->critical_section);
+		this->event = CreateEvent(NULL, FALSE, FALSE, NULL);
 	}
 
 	/* virtual */ ~ThreadMutex_Win32()
 	{
 		DeleteCriticalSection(&this->critical_section);
+		CloseHandle(this->event);
 	}
 
 	/* virtual */ void BeginCritical()
@@ -128,6 +131,18 @@ public:
 	{
 		LeaveCriticalSection(&this->critical_section);
 	}
+
+	/* virtual */ void WaitForSignal()
+	{
+		this->EndCritical();
+		WaitForSingleObject(this->event, INFINITE);
+		this->BeginCritical();
+	}
+
+	/* virtual */ void SendSignal()
+	{
+		SetEvent(this->event);
+	}
 };
 
 /* static */ ThreadMutex *ThreadMutex::New()