GPU: tiny optimization and log fix

GPU/GLES/DisplayListInterpreter.cpp

@@ -338,7 +338,7 @@ void GLES_GPU::FastRunLoop(DisplayList &list) {
 		u32 cmd = op >> 24;
 
 		u32 diff = op ^ gstate.cmdmem[cmd];
-		CheckFlushOp(op, diff);
+		CheckFlushOp(cmd, diff);
 		gstate.cmdmem[cmd] = op;
 		ExecuteOp(op, diff);
 
@@ -346,8 +346,7 @@ void GLES_GPU::FastRunLoop(DisplayList &list) {
 	}
 }
 
-inline void GLES_GPU::CheckFlushOp(u32 op, u32 diff) {
-	u32 cmd = op >> 24;
+inline void GLES_GPU::CheckFlushOp(int cmd, u32 diff) {
 	if (flushBeforeCommand_[cmd] == 1 || (diff && flushBeforeCommand_[cmd] == 2))
 	{
 		if (dumpThisFrame_) {
@@ -358,7 +357,7 @@ inline void GLES_GPU::CheckFlushOp(u32 op, u32 diff) {
 }
 
 void GLES_GPU::PreExecuteOp(u32 op, u32 diff) {
-	CheckFlushOp(op, diff);
+	CheckFlushOp(op >> 24, diff);
 }
 
 void GLES_GPU::ExecuteOp(u32 op, u32 diff) {

GPU/GLES/DisplayListInterpreter.h

@@ -73,7 +73,7 @@ protected:
 private:
 	void DoBlockTransfer();
 	void ApplyDrawState(int prim);
-	void CheckFlushOp(u32 op, u32 diff);
+	void CheckFlushOp(int cmd, u32 diff);
 	void BuildReportingInfo();
 
 	FramebufferManager framebufferManager_;

GPU/GLES/TransformPipeline.cpp

@@ -1057,6 +1057,7 @@ u32 TransformDrawEngine::ComputeHash() {
 	int vertexSize = dec_->GetDecVtxFmt().stride;
 
 	// TODO: Add some caps both for numDrawCalls and num verts to check?
+	// It is really very expensive to check all the vertex data so often.
 	for (int i = 0; i < numDrawCalls; i++) {
 		if (!drawCalls[i].inds) {
 			fullhash += CityHash32((const char *)drawCalls[i].verts, vertexSize * drawCalls[i].vertexCount);

GPU/GeDisasm.cpp

@@ -1074,7 +1074,7 @@ void GeDisassembleOp(u32 pc, u32 op, u32 prev, char *buffer) {
 		if (data & ~0x7F)
 			sprintf(buffer, "BONE #%i (extra %x)", data & 0x7F, data);
 		else
-			sprintf(buffer, "BONE #%i", data & 0xF);
+			sprintf(buffer, "BONE #%i", data & 0x7F);
 		break;
 
 	case GE_CMD_BONEMATRIXDATA: