use multiple CPU threads to compile and selecting a dedicated GPU if possible

src/lve_device.cpp

@@ -119,7 +119,14 @@ void LveDevice::pickPhysicalDevice() {
   vkEnumeratePhysicalDevices(instance, &deviceCount, devices.data());
 
   for (const auto &device : devices) {
-    if (isDeviceSuitable(device)) {
+    if (isPreferredDevice(device)) {
+      physicalDevice = device;
+      return;
+    }
+  }
+
+  for (const auto &device : devices) {
+    if (isSuitableDevice(device)) {
       physicalDevice = device;
       break;
     }
@@ -195,7 +202,19 @@ void LveDevice::createCommandPool() {
 
 void LveDevice::createSurface() { window.createWindowSurface(instance, &surface_); }
 
-bool LveDevice::isDeviceSuitable(VkPhysicalDevice device) {
+bool LveDevice::isPreferredDevice(VkPhysicalDevice device) {
+  if (!isSuitableDevice(device))
+  {
+      return false;
+  }
+
+  auto props = VkPhysicalDeviceProperties{};
+  vkGetPhysicalDeviceProperties(device, &props);
+
+  return props.deviceType == VkPhysicalDeviceType::VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU;
+}
+
+bool LveDevice::isSuitableDevice(VkPhysicalDevice device) {
   QueueFamilyIndices indices = findQueueFamilies(device);
 
   bool extensionsSupported = checkDeviceExtensionSupport(device);
@@ -530,4 +549,4 @@ void LveDevice::createImageWithInfo(
   }
 }
 
-}  // namespace lve
+}  // namespace lve

src/lve_device.hpp

@@ -81,7 +81,8 @@ class LveDevice {
   void createCommandPool();
 
   // helper functions
-  bool isDeviceSuitable(VkPhysicalDevice device);
+  bool isSuitableDevice(VkPhysicalDevice device);
+  bool isPreferredDevice(VkPhysicalDevice device);
   std::vector<const char *> getRequiredExtensions();
   bool checkValidationLayerSupport();
   QueueFamilyIndices findQueueFamilies(VkPhysicalDevice device);

unixBuild.sh

@@ -2,5 +2,10 @@
 mkdir -p build
 cd build
 cmake -S ../ -B .
+if [ -z "$MAKEFLAGS" ]
+then
+      export MAKEFLAGS=-j$(nproc)
+fi
+
 make && make Shaders && ./LveEngine
-cd ..
+cd ..