[GPU] Vsync timing requires far less cpu now to be accurate

src/xenia/gpu/graphics_system.cc

@@ -104,22 +104,34 @@ X_STATUS GraphicsSystem::Setup(cpu::Processor* processor,
   vsync_worker_running_ = true;
   vsync_worker_thread_ = kernel::object_ref<kernel::XHostThread>(
       new kernel::XHostThread(kernel_state_, 128 * 1024, 0, [this]() {
-        double vsync_duration_d =
+        const double vsync_duration_d =
             cvars::vsync
                 ? std::max<double>(
                       5.0, 1000.0 / static_cast<double>(cvars::vsync_fps))
                 : 1.0;
         uint64_t last_frame_time = Clock::QueryGuestTickCount();
-        while (vsync_worker_running_) {
+        // Sleep for 90% of the vblank duration, spin for 10%
-          uint64_t current_time = Clock::QueryGuestTickCount();
+        const double duration_scalar = 0.90;
 
-          uint64_t tick_freq = Clock::guest_tick_frequency();
+        while (vsync_worker_running_) {
-          uint64_t time_delta = current_time - last_frame_time;
+          const uint64_t current_time = Clock::QueryGuestTickCount();
-          double elapsed_d = static_cast<double>(time_delta) /
+          const uint64_t tick_freq = Clock::guest_tick_frequency();
+          const uint64_t time_delta = current_time - last_frame_time;
+          const double elapsed_d = static_cast<double>(time_delta) /
                                    (static_cast<double>(tick_freq) / 1000.0);
           if (elapsed_d >= vsync_duration_d) {
-            MarkVblank();
             last_frame_time = current_time;
+
+            // TODO(disjtqz): should recalculate the remaining time to a vblank
+            // after MarkVblank, no idea how long the guest code normally takes
+            MarkVblank();
+            if (cvars::vsync) {
+              const uint64_t estimated_nanoseconds = static_cast<uint64_t>(
+                  (vsync_duration_d * 1000000.0) *
+                  duration_scalar);  // 1000 microseconds = 1 ms
+
+              threading::NanoSleep(estimated_nanoseconds);
+            }
           }
           if (!cvars::vsync) {
             xe::threading::Sleep(std::chrono::milliseconds(1));
@@ -131,7 +143,8 @@ X_STATUS GraphicsSystem::Setup(cpu::Processor* processor,
   vsync_worker_thread_->set_can_debugger_suspend(true);
   vsync_worker_thread_->set_name("GPU VSync");
   vsync_worker_thread_->Create();
-  vsync_worker_thread_->thread()->set_priority(threading::ThreadPriority::kLowest);
+  vsync_worker_thread_->thread()->set_priority(
+      threading::ThreadPriority::kLowest);
   if (cvars::trace_gpu_stream) {
     BeginTracing();
   }