Merge pull request #18598 from hrydgard/inline-update-uvscale

Inline UpdateUVScaleOffset
hrydgard · Dec 21, 2023 · 03b8826 · 03b8826
2 parents 5fccf64 + e488189
commit 03b8826
Show file tree

Hide file tree

Showing 5 changed files with 39 additions and 27 deletions.
diff --git a/GPU/GPUCommon.cpp b/GPU/GPUCommon.cpp
@@ -1287,7 +1287,7 @@ void GPUCommon::FlushImm() {
 		immCount_ = 0;
 		return;
 	}
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	if (vfb) {
 		CheckDepthUsage(vfb);
 	}
@@ -1933,7 +1933,7 @@ bool GPUCommon::PerformWriteStencilFromMemory(u32 dest, int size, WriteStencil f
 }
 
 bool GPUCommon::GetCurrentSimpleVertices(int count, std::vector<GPUDebugVertex> &vertices, std::vector<u16> &indices) {
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	return drawEngineCommon_->GetCurrentSimpleVertices(count, vertices, indices);
 }
 
@@ -1942,18 +1942,3 @@ bool GPUCommon::DescribeCodePtr(const u8 *ptr, std::string &name) {
 	// which is owned by the drawengine.
 	return drawEngineCommon_->DescribeCodePtr(ptr, name);
 }
-
-void GPUCommon::UpdateUVScaleOffset() {
-#if defined(_M_SSE)
-	__m128i values = _mm_slli_epi32(_mm_load_si128((const __m128i *)&gstate.texscaleu), 8);
-	_mm_storeu_si128((__m128i *)&gstate_c.uv, values);
-#elif PPSSPP_ARCH(ARM_NEON)
-	const uint32x4_t values = vshlq_n_u32(vld1q_u32((const u32 *)&gstate.texscaleu), 8);
-	vst1q_u32((u32 *)&gstate_c.uv, values);
-#else
-	gstate_c.uv.uScale = getFloat24(gstate.texscaleu);
-	gstate_c.uv.vScale = getFloat24(gstate.texscalev);
-	gstate_c.uv.uOff = getFloat24(gstate.texoffsetu);
-	gstate_c.uv.vOff = getFloat24(gstate.texoffsetv);
-#endif
-}
diff --git a/GPU/GPUCommon.h b/GPU/GPUCommon.h
@@ -204,8 +204,6 @@ class GPUCommon : public GPUInterface, public GPUDebugInterface {
 	GPUgstate GetGState() override;
 	void SetCmdValue(u32 op) override;
 
-	void UpdateUVScaleOffset();
-
 	DisplayList* getList(int listid) override {
 		return &dls[listid];
 	}

diff --git a/GPU/GPUCommonHW.cpp b/GPU/GPUCommonHW.cpp
@@ -982,7 +982,7 @@ void GPUCommonHW::Execute_Prim(u32 op, u32 diff) {
 	}
 
 	int bytesRead = 0;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 
 	// cull mode
 	int cullMode = gstate.getCullMode();
@@ -1306,7 +1306,7 @@ void GPUCommonHW::Execute_Bezier(u32 op, u32 diff) {
 	}
 
 	int bytesRead = 0;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	drawEngineCommon_->SubmitCurve(control_points, indices, surface, gstate.vertType, &bytesRead, "bezier");
 
 	gstate_c.Dirty(DIRTY_RASTER_STATE | DIRTY_VERTEXSHADER_STATE | DIRTY_GEOMETRYSHADER_STATE | DIRTY_UVSCALEOFFSET);
@@ -1380,7 +1380,7 @@ void GPUCommonHW::Execute_Spline(u32 op, u32 diff) {
 	}
 
 	int bytesRead = 0;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	drawEngineCommon_->SubmitCurve(control_points, indices, surface, gstate.vertType, &bytesRead, "spline");
 
 	gstate_c.Dirty(DIRTY_RASTER_STATE | DIRTY_VERTEXSHADER_STATE | DIRTY_GEOMETRYSHADER_STATE | DIRTY_UVSCALEOFFSET);

diff --git a/GPU/GPUState.h b/GPU/GPUState.h
@@ -17,12 +17,25 @@
 
 #pragma once
 
+#include "ppsspp_config.h"
+
 #include "Common/CommonTypes.h"
 #include "Common/Swap.h"
 #include "GPU/GPU.h"
 #include "GPU/ge_constants.h"
 #include "GPU/Common/ShaderCommon.h"
 
+#if defined(_M_SSE)
+#include <emmintrin.h>
+#endif
+#if PPSSPP_ARCH(ARM_NEON)
+#if defined(_MSC_VER) && PPSSPP_ARCH(ARM64)
+#include <arm64_neon.h>
+#else
+#include <arm_neon.h>
+#endif
+#endif
+
 class PointerWrap;
 
 struct GPUgstate {
@@ -523,6 +536,8 @@ enum class SubmitType {
 	HW_SPLINE,
 };
 
+extern GPUgstate gstate;
+
 struct GPUStateCache {
 	bool Use(u32 flags) const { return (useFlags_ & flags) != 0; } // Return true if ANY of flags are true.
 	bool UseAll(u32 flags) const { return (useFlags_ & flags) == flags; } // Return true if ALL flags are true.
@@ -604,6 +619,21 @@ struct GPUStateCache {
 		return useFlags_;
 	}
 
+	void UpdateUVScaleOffset() {
+#if defined(_M_SSE)
+		__m128i values = _mm_slli_epi32(_mm_load_si128((const __m128i *)&gstate.texscaleu), 8);
+		_mm_storeu_si128((__m128i *)&uv, values);
+#elif PPSSPP_ARCH(ARM_NEON)
+		const uint32x4_t values = vshlq_n_u32(vld1q_u32((const u32 *)&gstate.texscaleu), 8);
+		vst1q_u32((u32 *)&uv, values);
+#else
+		uv.uScale = getFloat24(gstate.texscaleu);
+		uv.vScale = getFloat24(gstate.texscalev);
+		uv.uOff = getFloat24(gstate.texoffsetu);
+		uv.vOff = getFloat24(gstate.texoffsetv);
+#endif
+	}
+
 private:
 	u32 useFlags_;
 public:
@@ -690,7 +720,6 @@ struct GPUStateCache {
 class GPUInterface;
 class GPUDebugInterface;
 
-extern GPUgstate gstate;
 extern GPUStateCache gstate_c;
 
 inline u32 GPUStateCache::getRelativeAddress(u32 data) const {

diff --git a/GPU/Software/SoftGpu.cpp b/GPU/Software/SoftGpu.cpp
@@ -864,7 +864,7 @@ void SoftGPU::Execute_Prim(u32 op, u32 diff) {
 
 	cyclesExecuted += EstimatePerVertexCost() * count;
 	int bytesRead;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	drawEngine_->transformUnit.SetDirty(dirtyFlags_);
 	drawEngine_->transformUnit.SubmitPrimitive(verts, indices, prim, count, gstate.vertType, &bytesRead, drawEngine_);
 	dirtyFlags_ = drawEngine_->transformUnit.GetDirty();
@@ -917,7 +917,7 @@ void SoftGPU::Execute_Bezier(u32 op, u32 diff) {
 	SetDrawType(DRAW_BEZIER, PatchPrimToPrim(surface.primType));
 
 	int bytesRead = 0;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	drawEngine_->transformUnit.SetDirty(dirtyFlags_);
 	drawEngineCommon_->SubmitCurve(control_points, indices, surface, gstate.vertType, &bytesRead, "bezier");
 	dirtyFlags_ = drawEngine_->transformUnit.GetDirty();
@@ -971,7 +971,7 @@ void SoftGPU::Execute_Spline(u32 op, u32 diff) {
 	SetDrawType(DRAW_SPLINE, PatchPrimToPrim(surface.primType));
 
 	int bytesRead = 0;
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	drawEngine_->transformUnit.SetDirty(dirtyFlags_);
 	drawEngineCommon_->SubmitCurve(control_points, indices, surface, gstate.vertType, &bytesRead, "spline");
 	dirtyFlags_ = drawEngine_->transformUnit.GetDirty();
@@ -1445,7 +1445,7 @@ bool SoftGPU::GetCurrentClut(GPUDebugBuffer &buffer)
 }
 
 bool SoftGPU::GetCurrentSimpleVertices(int count, std::vector<GPUDebugVertex> &vertices, std::vector<u16> &indices) {
-	UpdateUVScaleOffset();
+	gstate_c.UpdateUVScaleOffset();
 	return drawEngine_->transformUnit.GetCurrentSimpleVertices(count, vertices, indices);
 }