OrdinaryMagician/vkdoom_m
OrdinaryMagician/vkdoom_m
1
Fork 0
Code Activity

- Replace subsector gbuffer in softpoly with a zbuffer

Browse source
This commit is contained in:
Magnus Norddahl 2017-08-12 15:16:31 +02:00
commit bf38fd57b0
21 changed files with 223 additions and 167 deletions
Download patch file Download diff file Expand all files Collapse all files

252
src/polyrenderer/drawers/screen_triangle.cpp
View file

@ -76,10 +76,9 @@ private:
int clipright;
int clipbottom;
// Subsector buffer
uint32_t * RESTRICT subsectorGBuffer;
uint32_t subsectorDepth;
int32_t subsectorPitch;
// Depth buffer
float * RESTRICT zbuffer;
int32_t zbufferPitch;
// Triangle bounding block
int minx, miny;
@ -113,10 +112,10 @@ private:
void CoverageTest();
void StencilEqualTest();
void StencilGreaterEqualTest();
void SubsectorTest();
void DepthTest(const TriDrawTriangleArgs *args);
void ClipTest();
void StencilWrite();
void SubsectorWrite();
void DepthWrite(const TriDrawTriangleArgs *args);
};
TriangleBlock::TriangleBlock(const TriDrawTriangleArgs *args)
@ -134,9 +133,8 @@ TriangleBlock::TriangleBlock(const TriDrawTriangleArgs *args)
stencilTestValue = args->uniforms->StencilTestValue();
stencilWriteValue = args->uniforms->StencilWriteValue();
subsectorGBuffer = args->subsectorGBuffer;
subsectorDepth = args->uniforms->SubsectorDepth();
subsectorPitch = args->stencilPitch;
zbuffer = args->zbuffer;
zbufferPitch = args->stencilPitch;
// 28.4 fixed-point coordinates
#ifdef NO_SSE
@ -235,10 +233,10 @@ void TriangleBlock::Loop(const TriDrawTriangleArgs *args, WorkerThreadData *thre
int core_skip = (num_cores - ((miny / q) - core) % num_cores) % num_cores;
int start_miny = miny + core_skip * q;
bool subsectorTest = args->uniforms->SubsectorTest();
bool depthTest = args->uniforms->DepthTest();
bool writeColor = args->uniforms->WriteColor();
bool writeStencil = args->uniforms->WriteStencil();
bool writeSubsector = args->uniforms->WriteSubsector();
bool writeDepth = args->uniforms->WriteDepth();
int bmode = (int)args->uniforms->BlendMode();
auto drawFunc = args->destBgra ? ScreenTriangle::TriDrawers32[bmode] : ScreenTriangle::TriDrawers8[bmode];
@ -259,8 +257,8 @@ void TriangleBlock::Loop(const TriDrawTriangleArgs *args, WorkerThreadData *thre
if (Mask0 == 0 && Mask1 == 0)
continue;
// To do: make the stencil test use its own flag for comparison mode instead of abusing the subsector test..
if (!subsectorTest)
// To do: make the stencil test use its own flag for comparison mode instead of abusing the depth test..
if (!depthTest)
{
StencilEqualTest();
if (Mask0 == 0 && Mask1 == 0)
@ -272,7 +270,7 @@ void TriangleBlock::Loop(const TriDrawTriangleArgs *args, WorkerThreadData *thre
if (Mask0 == 0 && Mask1 == 0)
continue;
SubsectorTest();
DepthTest(args);
if (Mask0 == 0 && Mask1 == 0)
continue;
}
@ -281,34 +279,54 @@ void TriangleBlock::Loop(const TriDrawTriangleArgs *args, WorkerThreadData *thre
drawFunc(X, Y, Mask0, Mask1, args);
if (writeStencil)
StencilWrite();
if (writeSubsector)
SubsectorWrite();
if (writeDepth)
DepthWrite(args);
}
}
}
#ifdef NO_SSE
void TriangleBlock::SubsectorTest()
void TriangleBlock::DepthTest(const TriDrawTriangleArgs *args)
{
int block = (X >> 3) + (Y >> 3) * subsectorPitch;
uint32_t *subsector = subsectorGBuffer + block * 64;
int block = (X >> 3) + (Y >> 3) * zbufferPitch;
float *depth = zbuffer + block * 64;
const TriVertex &v1 = *args->v1;
float stepXW = args->gradientX.W;
float stepYW = args->gradientY.W;
float posYW = v1.w + stepXW * (X - v1.x) + stepYW * (Y - v1.y);
uint32_t mask0 = 0;
uint32_t mask1 = 0;
for (int i = 0; i < 32; i++)
for (int iy = 0; iy < 4; iy++)
{
bool covered = *subsector >= subsectorDepth;
mask0 <<= 1;
mask0 |= (uint32_t)covered;
subsector++;
float posXW = posYW;
for (int ix = 0; ix < 8; ix++)
{
bool covered = *depth <= posXW;
mask0 <<= 1;
mask0 |= (uint32_t)covered;
depth++;
posXW += stepXW;
}
posYW += stepYW;
}
for (int i = 0; i < 32; i++)
for (int iy = 0; iy < 4; iy++)
{
bool covered = *subsector >= subsectorDepth;
mask1 <<= 1;
mask1 |= (uint32_t)covered;
subsector++;
float posXW = posYW;
for (int ix = 0; ix < 8; ix++)
{
bool covered = *depth <= posXW;
mask1 <<= 1;
mask1 |= (uint32_t)covered;
depth++;
posXW += stepXW;
}
posYW += stepYW;
}
Mask0 = Mask0 & mask0;
@ -317,26 +335,50 @@ void TriangleBlock::SubsectorTest()
#else
void TriangleBlock::SubsectorTest()
void TriangleBlock::DepthTest(const TriDrawTriangleArgs *args)
{
int block = (X >> 3) + (Y >> 3) * subsectorPitch;
uint32_t *subsector = subsectorGBuffer + block * 64;
int block = (X >> 3) + (Y >> 3) * zbufferPitch;
float *depth = zbuffer + block * 64;
const TriVertex &v1 = *args->v1;
float stepXW = args->gradientX.W;
float stepYW = args->gradientY.W;
float posYW = v1.w + stepXW * (X - v1.x) + stepYW * (Y - v1.y);
__m128 mposYW = _mm_setr_ps(posYW, posYW + stepXW, posYW + stepXW + stepXW, posYW + stepXW + stepXW + stepXW);
__m128 mstepXW = _mm_set1_ps(stepXW * 4.0f);
__m128 mstepYW = _mm_set1_ps(stepYW);
uint32_t mask0 = 0;
uint32_t mask1 = 0;
__m128i msubsectorDepth = _mm_set1_epi32(subsectorDepth);
__m128i mnotxor = _mm_set1_epi32(0xffffffff);
for (int iy = 0; iy < 8; iy++)
for (int iy = 0; iy < 4; iy++)
{
mask0 <<= 4;
mask0 |= _mm_movemask_ps(_mm_castsi128_ps(_mm_shuffle_epi32(_mm_xor_si128(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)subsector), msubsectorDepth), mnotxor), _MM_SHUFFLE(0, 1, 2, 3))));
subsector += 4;
__m128 mposXW = mposYW;
for (int ix = 0; ix < 2; ix++)
{
__m128 covered = _mm_cmplt_ps(_mm_loadu_ps(depth), mposXW);
mask0 <<= 4;
mask0 |= _mm_movemask_ps(_mm_shuffle_ps(covered, covered, _MM_SHUFFLE(0, 1, 2, 3)));
depth += 4;
mposXW = _mm_add_ps(mposXW, mstepXW);
}
mposYW = _mm_add_ps(mposYW, mstepYW);
}
for (int iy = 0; iy < 8; iy++)
for (int iy = 0; iy < 4; iy++)
{
mask1 <<= 4;
mask1 |= _mm_movemask_ps(_mm_castsi128_ps(_mm_shuffle_epi32(_mm_xor_si128(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)subsector), msubsectorDepth), mnotxor), _MM_SHUFFLE(0, 1, 2, 3))));
subsector += 4;
__m128 mposXW = mposYW;
for (int ix = 0; ix < 2; ix++)
{
__m128 covered = _mm_cmplt_ps(_mm_loadu_ps(depth), mposXW);
mask1 <<= 4;
mask1 |= _mm_movemask_ps(_mm_shuffle_ps(covered, covered, _MM_SHUFFLE(0, 1, 2, 3)));
depth += 4;
mposXW = _mm_add_ps(mposXW, mstepXW);
}
mposYW = _mm_add_ps(mposYW, mstepYW);
}
Mask0 = Mask0 & mask0;
@ -798,65 +840,91 @@ void TriangleBlock::StencilWrite()
#ifdef NO_SSE
void TriangleBlock::SubsectorWrite()
void TriangleBlock::DepthWrite(const TriDrawTriangleArgs *args)
{
int block = (X >> 3) + (Y >> 3) * subsectorPitch;
uint32_t *subsector = subsectorGBuffer + block * 64;
int block = (X >> 3) + (Y >> 3) * zbufferPitch;
float *depth = zbuffer + block * 64;
const TriVertex &v1 = *args->v1;
float stepXW = args->gradientX.W;
float stepYW = args->gradientY.W;
float posYW = v1.w + stepXW * (X - v1.x) + stepYW * (Y - v1.y);
if (Mask0 == 0xffffffff && Mask1 == 0xffffffff)
{
for (int i = 0; i < 64; i++)
for (int iy = 0; iy < 8; iy++)
{
*(subsector++) = subsectorDepth;
float posXW = posYW;
for (int ix = 0; ix < 8; ix++)
{
*(depth++) = posXW;
posXW += stepXW;
}
posYW += stepYW;
}
}
else
{
uint32_t mask0 = Mask0;
uint32_t mask1 = Mask1;
for (int i = 0; i < 32; i++)
for (int iy = 0; iy < 4; iy++)
{
if (mask0 & (1 << 31))
*subsector = subsectorDepth;
mask0 <<= 1;
subsector++;
float posXW = posYW;
for (int ix = 0; ix < 8; ix++)
{
if (mask0 & (1 << 31))
*depth = posXW;
posXW += stepXW;
mask0 <<= 1;
depth++;
}
posYW += stepYW;
}
for (int i = 0; i < 32; i++)
for (int iy = 0; iy < 4; iy++)
{
if (mask1 & (1 << 31))
*subsector = subsectorDepth;
mask1 <<= 1;
subsector++;
float posXW = posYW;
for (int ix = 0; ix < 8; ix++)
{
if (mask1 & (1 << 31))
*depth = posXW;
posXW += stepXW;
mask1 <<= 1;
depth++;
}
posYW += stepYW;
}
}
}
#else
void TriangleBlock::SubsectorWrite()
void TriangleBlock::DepthWrite(const TriDrawTriangleArgs *args)
{
int block = (X >> 3) + (Y >> 3) * subsectorPitch;
uint32_t *subsector = subsectorGBuffer + block * 64;
__m128i msubsectorDepth = _mm_set1_epi32(subsectorDepth);
int block = (X >> 3) + (Y >> 3) * zbufferPitch;
float *depth = zbuffer + block * 64;
const TriVertex &v1 = *args->v1;
float stepXW = args->gradientX.W;
float stepYW = args->gradientY.W;
float posYW = v1.w + stepXW * (X - v1.x) + stepYW * (Y - v1.y);
__m128 mposYW = _mm_setr_ps(posYW, posYW + stepXW, posYW + stepXW + stepXW, posYW + stepXW + stepXW + stepXW);
__m128 mstepXW = _mm_set1_ps(stepXW * 4.0f);
__m128 mstepYW = _mm_set1_ps(stepYW);
if (Mask0 == 0xffffffff && Mask1 == 0xffffffff)
{
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth); subsector += 4;
_mm_storeu_si128((__m128i*)subsector, msubsectorDepth);
for (int iy = 0; iy < 8; iy++)
{
__m128 mposXW = mposYW;
_mm_storeu_ps(depth, mposXW); depth += 4; mposXW = _mm_add_ps(mposXW, mstepXW);
_mm_storeu_ps(depth, mposXW); depth += 4;
mposYW = _mm_add_ps(mposYW, mstepYW);
}
}
else
{
@ -866,23 +934,21 @@ void TriangleBlock::SubsectorWrite()
__m128i mmask0 = _mm_set1_epi32(Mask0);
__m128i mmask1 = _mm_set1_epi32(Mask1);
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask0 = _mm_slli_epi32(mmask0, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); subsector += 4;
for (int iy = 0; iy < 4; iy++)
{
__m128 mposXW = mposYW;
_mm_maskmoveu_si128(_mm_castps_si128(mposXW), _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)depth); mmask0 = _mm_slli_epi32(mmask0, 4); depth += 4; mposXW = _mm_add_ps(mposXW, mstepXW);
_mm_maskmoveu_si128(_mm_castps_si128(mposXW), _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask0, topfour), _mm_setzero_si128()), mxormask), (char*)depth); mmask0 = _mm_slli_epi32(mmask0, 4); depth += 4;
mposYW = _mm_add_ps(mposYW, mstepYW);
}
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); mmask1 = _mm_slli_epi32(mmask1, 4); subsector += 4;
_mm_maskmoveu_si128(msubsectorDepth, _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)subsector); subsector += 4;
for (int iy = 0; iy < 4; iy++)
{
__m128 mposXW = mposYW;
_mm_maskmoveu_si128(_mm_castps_si128(mposXW), _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)depth); mmask1 = _mm_slli_epi32(mmask1, 4); depth += 4; mposXW = _mm_add_ps(mposXW, mstepXW);
_mm_maskmoveu_si128(_mm_castps_si128(mposXW), _mm_xor_si128(_mm_cmpeq_epi32(_mm_and_si128(mmask1, topfour), _mm_setzero_si128()), mxormask), (char*)depth); mmask1 = _mm_slli_epi32(mmask1, 4); depth += 4;
mposYW = _mm_add_ps(mposYW, mstepYW);
}
}
}