Ensure stuff run on every prim are inlined.

These both get a ton of calls and show up on profiles.

GPU/GLES/Framebuffer.cpp

@@ -660,7 +660,7 @@ void FramebufferManager::DestroyFramebuf(VirtualFramebuffer *v) {
 	delete v;
 }
 
-void FramebufferManager::SetRenderFrameBuffer() {
+void FramebufferManager::DoSetRenderFrameBuffer() {
 	if (!gstate_c.framebufChanged && currentRenderVfb_) {
 		currentRenderVfb_->last_frame_render = gpuStats.numFlips;
 		currentRenderVfb_->dirtyAfterDisplay = true;

GPU/GLES/Framebuffer.h

@@ -135,7 +135,18 @@ public:
 	void Resized();
 	void DeviceLost();
 	void CopyDisplayToOutput();
-	void SetRenderFrameBuffer();  // Uses parameters computed from gstate
+	void DoSetRenderFrameBuffer();  // Uses parameters computed from gstate
+	void SetRenderFrameBuffer() {
+		// Inlining this part since it's so frequent.
+		if (!gstate_c.framebufChanged && currentRenderVfb_) {
+			currentRenderVfb_->last_frame_render = gpuStats.numFlips;
+			currentRenderVfb_->dirtyAfterDisplay = true;
+			if (!gstate_c.skipDrawReason)
+				currentRenderVfb_->reallyDirtyAfterDisplay = true;
+			return;
+		}
+		DoSetRenderFrameBuffer();
+	}
 	void UpdateFromMemory(u32 addr, int size, bool safe);
 	void SetLineWidth();
 

GPU/GLES/TransformPipeline.cpp

@@ -267,35 +267,6 @@ void TransformDrawEngine::SetupVertexDecoder(u32 vertType) {
 	}
 }
 
-int TransformDrawEngine::EstimatePerVertexCost() {
-	// TODO: This is transform cost, also account for rasterization cost somehow... although it probably
-	// runs in parallel with transform.
-
-	// Also, this is all pure guesswork. If we can find a way to do measurements, that would be great.
-
-	// GTA wants a low value to run smooth, GoW wants a high value (otherwise it thinks things
-	// went too fast and starts doing all the work over again).
-
-	int cost = 20;
-	if (gstate.isLightingEnabled()) {
-		cost += 10;
-
-		for (int i = 0; i < 4; i++) {
-			if (gstate.isLightChanEnabled(i))
-				cost += 10;
-		}
-	}
-
-	if (gstate.getUVGenMode() != GE_TEXMAP_TEXTURE_COORDS) {
-		cost += 20;
-	}
-	if (dec_ && dec_->morphcount > 1) {
-		cost += 5 * dec_->morphcount;
-	}
-
-	return cost;
-}
-
 void TransformDrawEngine::SubmitPrim(void *verts, void *inds, GEPrimitiveType prim, int vertexCount, u32 vertType, int *bytesRead) {
 	if (vertexCount == 0)
 		return;  // we ignore zero-sized draw calls.

GPU/GLES/TransformPipeline.h

@@ -21,6 +21,7 @@
 
 #include "GPU/Common/GPUDebugInterface.h"
 #include "GPU/Common/IndexGenerator.h"
+#include "GPU/GLES/VertexDecoder.h"
 #include "gfx/gl_common.h"
 #include "gfx/gl_lost_manager.h"
 
@@ -28,8 +29,6 @@ class LinkedShader;
 class ShaderManager;
 class TextureCache;
 class FramebufferManager;
-class VertexDecoder;
-class VertexDecoderJitCache;
 struct TransformedVertex;
 
 struct DecVtxFormat;
@@ -127,7 +126,34 @@ public:
 	void SetupVertexDecoder(u32 vertType);
 
 	// This requires a SetupVertexDecoder call first.
-	int EstimatePerVertexCost();
+	int EstimatePerVertexCost() {
+		// TODO: This is transform cost, also account for rasterization cost somehow... although it probably
+		// runs in parallel with transform.
+
+		// Also, this is all pure guesswork. If we can find a way to do measurements, that would be great.
+
+		// GTA wants a low value to run smooth, GoW wants a high value (otherwise it thinks things
+		// went too fast and starts doing all the work over again).
+
+		int cost = 20;
+		if (gstate.isLightingEnabled()) {
+			cost += 10;
+
+			for (int i = 0; i < 4; i++) {
+				if (gstate.isLightChanEnabled(i))
+					cost += 10;
+			}
+		}
+
+		if (gstate.getUVGenMode() != GE_TEXMAP_TEXTURE_COORDS) {
+			cost += 20;
+		}
+		if (dec_ && dec_->morphcount > 1) {
+			cost += 5 * dec_->morphcount;
+		}
+
+		return cost;
+	}
 
 	// So that this can be inlined
 	void Flush() {