Emulator-Archive/ppsspp
1
0
Fork 0
mirror of https://github.com/hrydgard/ppsspp.git synced 2024-11-23 21:39:52 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Some very minor optimizations. Remove little-used stat counter.

Browse Source
This commit is contained in:
Henrik Rydgard 2014-10-19 20:25:04 +02:00
parent 87bcb56e95
commit 7cb5e7f53f
6 changed files with 49 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
GPU/Common/VertexDecoderCommon.cpp
View File

@ -134,8 +134,7 @@ void PrintDecodedVertex(VertexReader &vtx) {
printf("P: %f %f %f\n", pos[0], pos[1], pos[2]);
}
VertexDecoder::VertexDecoder() : coloff(0), nrmoff(0), posoff(0), jitted_(0) {
memset(stats_, 0, sizeof(stats_));
VertexDecoder::VertexDecoder() : jitted_(0) {
}
void VertexDecoder::Step_WeightsU8() const
@ -1090,7 +1089,6 @@ int VertexDecoder::ToString(char *output) const {
output += sprintf(output, "I: %i ", idx);
if (morphcount > 1)
output += sprintf(output, "Morph: %i ", morphcount);
output += sprintf(output, "Verts: %i ", stats_[STAT_VERTSSUBMITTED]);
if (throughmode)
output += sprintf(output, " (through)");

52
GPU/Common/VertexDecoderCommon.h
View File

@ -85,11 +85,6 @@ struct TransformedVertex
void GetIndexBounds(const void *inds, int count, u32 vertType, u16 *indexLowerBound, u16 *indexUpperBound);
enum {
STAT_VERTSSUBMITTED = 0,
NUM_VERTEX_DECODER_STATS = 1
};
inline int RoundUp4(int x) {
return (x + 3) & ~3;
}
@ -524,14 +519,6 @@ public:
void Step_PosS16Through() const;
void Step_PosFloatThrough() const;
void ResetStats() {
memset(stats_, 0, sizeof(stats_));
}
void IncrementStat(int stat, int amount) {
stats_[stat] += amount;
}
// output must be big for safety.
// Returns number of chars written.
// Ugly for speed.
@ -541,9 +528,11 @@ public:
mutable u8 *decoded_;
mutable const u8 *ptr_;
JittedVertexDecoder jitted_;
// "Immutable" state, set at startup
// The decoding steps
// The decoding steps. Never more than 5.
StepFunction steps_[5];
int numSteps_;
@ -551,28 +540,23 @@ public:
DecVtxFormat decFmt;
bool throughmode;
int biggest;
int size;
int onesize_;
u8 size;
u8 onesize_;
int weightoff;
int tcoff;
int coloff;
int nrmoff;
int posoff;
u8 weightoff;
u8 tcoff;
u8 coloff;
u8 nrmoff;
u8 posoff;
int tc;
int col;
int nrm;
int pos;
int weighttype;
int idx;
int morphcount;
int nweights;
int stats_[NUM_VERTEX_DECODER_STATS];
JittedVertexDecoder jitted_;
u8 tc;
u8 col;
u8 nrm;
u8 pos;
u8 weighttype;
u8 idx;
u8 morphcount;
u8 nweights;
friend class VertexDecoderJitCache;
};

6
GPU/Directx9/SplineDX9.cpp
View File

@ -98,7 +98,8 @@ void TransformDrawEngineDX9::SubmitSpline(void* control_points, void* indices, i
gstate_c.uv.uOff = 0;
gstate_c.uv.vOff = 0;
}
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, 0);
int bytesRead;
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, &bytesRead);
Flush();
@ -190,7 +191,8 @@ void TransformDrawEngineDX9::SubmitBezier(void* control_points, void* indices, i
gstate_c.uv.vOff = 0;
}
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, 0);
int bytesRead;
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, &bytesRead);
Flush();
if (g_Config.bPrescaleUV) {

28
GPU/Directx9/TransformPipelineDX9.cpp
View File

@ -293,30 +293,22 @@ inline void TransformDrawEngineDX9::SetupVertexDecoderInternal(u32 vertType) {
}
void TransformDrawEngineDX9::SubmitPrim(void *verts, void *inds, GEPrimitiveType prim, int vertexCount, u32 vertType, int *bytesRead) {
if (vertexCount == 0)
return; // we ignore zero-sized draw calls.
if (!indexGen.PrimCompatible(prevPrim_, prim) || numDrawCalls >= MAX_DEFERRED_DRAW_CALLS || vertexCountInDrawCalls + vertexCount > VERTEX_BUFFER_MAX)
Flush();
// TODO: Is this the right thing to do?
if (prim == GE_PRIM_KEEP_PREVIOUS) {
prim = prevPrim_;
}
prevPrim_ = prim;
if ((vertexCount < 2 && prim > 0) || (vertexCount < 3 && prim > 2 && prim != GE_PRIM_RECTANGLES))
return;
// TODO: Is this the right thing to do?
if (prim == GE_PRIM_KEEP_PREVIOUS) {
prim = prevPrim_;
} else {
prevPrim_ = prim;
}
SetupVertexDecoderInternal(vertType);
dec_->IncrementStat(STAT_VERTSSUBMITTED, vertexCount);
if (bytesRead)
*bytesRead = vertexCount * dec_->VertexSize();
gpuStats.numDrawCalls++;
gpuStats.numVertsSubmitted += vertexCount;
*bytesRead = vertexCount * dec_->VertexSize();
DeferredDrawCall &dc = drawCalls[numDrawCalls];
dc.verts = verts;
@ -358,6 +350,7 @@ void TransformDrawEngineDX9::SubmitPrim(void *verts, void *inds, GEPrimitiveType
}
if (prim == GE_PRIM_RECTANGLES && (gstate.getTextureAddress(0) & 0x3FFFFFFF) == (gstate.getFrameBufAddress() & 0x3FFFFFFF)) {
// Rendertarget == texture?
if (!g_Config.bDisableSlowFramebufEffects) {
gstate_c.textureChanged |= TEXCHANGE_PARAMSONLY;
Flush();
@ -888,6 +881,9 @@ rotateVBO:
}
}
gpuStats.numDrawCalls += numDrawCalls;
gpuStats.numVertsSubmitted += vertexCountInDrawCalls;
indexGen.Reset();
decodedVerts_ = 0;
numDrawCalls = 0;

6
GPU/GLES/Spline.cpp
View File

@ -100,7 +100,8 @@ void TransformDrawEngine::SubmitSpline(void* control_points, void* indices, int
gstate_c.uv.uOff = 0;
gstate_c.uv.vOff = 0;
}
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, 0);
int bytesRead;
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, &bytesRead);
Flush();
@ -192,7 +193,8 @@ void TransformDrawEngine::SubmitBezier(void* control_points, void* indices, int
gstate_c.uv.vOff = 0;
}
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, 0);
int bytesRead;
SubmitPrim(decoded2, quadIndices_, GE_PRIM_TRIANGLES, count, vertTypeWithIndex16, &bytesRead);
Flush();
if (g_Config.bPrescaleUV) {

21
GPU/GLES/TransformPipeline.cpp
View File

@ -275,27 +275,22 @@ inline void TransformDrawEngine::SetupVertexDecoderInternal(u32 vertType) {
}
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.
if (!indexGen.PrimCompatible(prevPrim_, prim) || numDrawCalls >= MAX_DEFERRED_DRAW_CALLS || vertexCountInDrawCalls + vertexCount > VERTEX_BUFFER_MAX)
Flush();
if ((vertexCount < 2 && prim > 0) || (vertexCount < 3 && prim > 2 && prim != GE_PRIM_RECTANGLES))
return;
// TODO: Is this the right thing to do?
if (prim == GE_PRIM_KEEP_PREVIOUS) {
prim = prevPrim_;
} else {
prevPrim_ = prim;
}
prevPrim_ = prim;
SetupVertexDecoderInternal(vertType);
dec_->IncrementStat(STAT_VERTSSUBMITTED, vertexCount);
if (bytesRead)
*bytesRead = vertexCount * dec_->VertexSize();
gpuStats.numDrawCalls++;
gpuStats.numVertsSubmitted += vertexCount;
*bytesRead = vertexCount * dec_->VertexSize();
DeferredDrawCall &dc = drawCalls[numDrawCalls];
dc.verts = verts;
@ -337,6 +332,7 @@ void TransformDrawEngine::SubmitPrim(void *verts, void *inds, GEPrimitiveType pr
}
if (prim == GE_PRIM_RECTANGLES && (gstate.getTextureAddress(0) & 0x3FFFFFFF) == (gstate.getFrameBufAddress() & 0x3FFFFFFF)) {
// Rendertarget == texture?
if (!g_Config.bDisableSlowFramebufEffects) {
gstate_c.textureChanged |= TEXCHANGE_PARAMSONLY;
Flush();
@ -902,6 +898,9 @@ rotateVBO:
}
}
gpuStats.numDrawCalls += numDrawCalls;
gpuStats.numVertsSubmitted += vertexCountInDrawCalls;
indexGen.Reset();
decodedVerts_ = 0;
numDrawCalls = 0;