zdoom/gzdoom
0
0
Fork 0
mirror of https://github.com/ZDoom/gzdoom.git synced 2024-11-29 15:22:08 +00:00
Code Issues Projects Releases Packages Wiki Activity

- Move depth testing out of span drawers

Browse source
This commit is contained in:
Magnus Norddahl 2017-11-28 23:08:28 +01:00
parent 01fd404133
commit fae514923a
1 changed files with 167 additions and 165 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

152
src/polyrenderer/drawers/screen_triangle.cpp
View file

@ -1179,21 +1179,88 @@ void ScreenTriangle::DrawSWRender(const TriDrawTriangleArgs *args, WorkerThreadD
// Draw the triangle: // Draw the triangle:
if (args->destBgra) auto drawfunc = (args->destBgra) ? DrawSpan32 : DrawSpan8;
{
float stepXW = args->gradientX.W;
float v1X = args->v1->x;
float v1Y = args->v1->y;
float v1W = args->v1->w;
int num_cores = thread->num_cores; int num_cores = thread->num_cores;
for (int y = topY + thread->skipped_by_thread(topY); y < bottomY; y += num_cores) for (int y = topY + thread->skipped_by_thread(topY); y < bottomY; y += num_cores)
{ {
DrawSpan32(y, leftEdge[y], rightEdge[y], args); int x = leftEdge[y];
} int xend = rightEdge[y];
}
else float *zbufferLine = args->zbuffer + args->stencilPitch * 8 * y;
float startX = x + (0.5f - v1X);
float startY = y + (0.5f - v1Y);
float posXW = v1W + stepXW * startX + args->gradientY.W * startY;
#ifndef NO_SSE
__m128 mstepXW = _mm_set1_ps(stepXW * 4.0f);
__m128 mfirstStepXW = _mm_setr_ps(0.0f, stepXW, stepXW + stepXW, stepXW + stepXW + stepXW);
while (x < xend)
{ {
int num_cores = thread->num_cores; int xstart = x;
for (int y = topY + thread->skipped_by_thread(topY); y < bottomY; y += num_cores)
int xendsse = x + ((xend - x) & ~3);
__m128 mposXW = _mm_add_ps(_mm_set1_ps(posXW), mfirstStepXW);
while (_mm_movemask_ps(_mm_cmple_ps(_mm_loadu_ps(zbufferLine + x), mposXW)) == 15 && x < xendsse)
{ {
DrawSpan8(y, leftEdge[y], rightEdge[y], args); _mm_storeu_ps(zbufferLine + x, mposXW);
mposXW = _mm_add_ps(mposXW, mstepXW);
x += 4;
} }
posXW = _mm_cvtss_f32(mposXW);
while (zbufferLine[x] <= posXW && x < xend)
{
zbufferLine[x] = posXW;
posXW += stepXW;
x++;
}
if (x > xstart)
drawfunc(y, xstart, x, args);
xendsse = x + ((xend - x) & ~3);
mposXW = _mm_add_ps(_mm_set1_ps(posXW), mfirstStepXW);
while (_mm_movemask_ps(_mm_cmple_ps(_mm_loadu_ps(zbufferLine + x), mposXW)) == 0 && x < xendsse)
{
mposXW = _mm_add_ps(mposXW, mstepXW);
x += 4;
}
posXW = _mm_cvtss_f32(mposXW);
while (zbufferLine[x] > posXW && x < xend)
{
posXW += stepXW;
x++;
}
}
#else
while (x < xend)
{
int xstart = x;
while (zbufferLine[x] <= posXW && x < xend)
{
zbufferLine[x] = posXW;
posXW += stepXW;
x++;
}
if (x > xstart)
drawfunc(y, xstart, x, args);
while (zbufferLine[x] > posXW && x < xend)
{
posXW += stepXW;
x++;
}
}
#endif
} }
} }
@ -1226,20 +1293,13 @@ void ScreenTriangle::DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f); float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += light >> 7; // 255 -> 256 light += light >> 7; // 255 -> 256
float *zbufferLine = args->zbuffer + args->stencilPitch * 8 * y;
uint32_t *dest = (uint32_t*)args->dest; uint32_t *dest = (uint32_t*)args->dest;
uint32_t *destLine = dest + args->pitch * y; uint32_t *destLine = dest + args->pitch * y;
int x = x0; int x = x0;
int sseEnd = x0 + (x1 - x0) / 2 * 2; int sseEnd = x0 + ((x1 - x0) & ~3);
while (x < sseEnd) while (x < sseEnd)
{ {
if (zbufferLine[x] <= posXW && zbufferLine[x + 1] <= posXW + stepXW)
{
zbufferLine[x] = posXW;
zbufferLine[x + 1] = posXW + stepXW;
uint32_t fgcolor[2]; uint32_t fgcolor[2];
int32_t lightshade[2]; int32_t lightshade[2];
@ -1288,51 +1348,8 @@ void ScreenTriangle::DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs
x += 2; x += 2;
} }
else
{
for (int i = 0; i < 2; i++)
{
if (zbufferLine[x] <= posXW)
{
float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW);
int32_t v = (int32_t)(posXV * rcpW);
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
uint32_t fgcolor = texPixels[texelX * texHeight + texelY];
uint32_t fgcolor_r = RPART(fgcolor);
uint32_t fgcolor_g = GPART(fgcolor);
uint32_t fgcolor_b = BPART(fgcolor);
uint32_t fgcolor_a = APART(fgcolor);
if (fgcolor_a > 127)
{
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * posXW), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
int lightshade = lightpos >> 8;
fgcolor_r = (fgcolor_r * lightshade) >> 8;
fgcolor_g = (fgcolor_g * lightshade) >> 8;
fgcolor_b = (fgcolor_b * lightshade) >> 8;
destLine[x] = 0xff000000 | (fgcolor_r << 16) | (fgcolor_g << 8) | fgcolor_b;
}
zbufferLine[x] = posXW;
}
posXW += stepXW;
posXU += stepXU;
posXV += stepXV;
x++;
}
}
}
while (x < x1) while (x < x1)
{
if (zbufferLine[x] <= posXW)
{ {
float rcpW = 0x01000000 / posXW; float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW); int32_t u = (int32_t)(posXU * rcpW);
@ -1359,9 +1376,6 @@ void ScreenTriangle::DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs
destLine[x] = 0xff000000 | (fgcolor_r << 16) | (fgcolor_g << 8) | fgcolor_b; destLine[x] = 0xff000000 | (fgcolor_r << 16) | (fgcolor_g << 8) | fgcolor_b;
} }
zbufferLine[x] = posXW;
}
posXW += stepXW; posXW += stepXW;
posXU += stepXU; posXU += stepXU;
posXV += stepXV; posXV += stepXV;
@ -1398,15 +1412,11 @@ void ScreenTriangle::DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f); float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += light >> 7; // 255 -> 256 light += light >> 7; // 255 -> 256
float *zbufferLine = args->zbuffer + args->stencilPitch * 8 * y;
uint32_t *dest = (uint32_t*)args->dest; uint32_t *dest = (uint32_t*)args->dest;
uint32_t *destLine = dest + args->pitch * y; uint32_t *destLine = dest + args->pitch * y;
int x = x0; int x = x0;
while (x < x1) while (x < x1)
{
if (zbufferLine[x] <= posXW)
{ {
float rcpW = 0x01000000 / posXW; float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW); int32_t u = (int32_t)(posXU * rcpW);
@ -1433,9 +1443,6 @@ void ScreenTriangle::DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs
destLine[x] = 0xff000000 | (fgcolor_r << 16) | (fgcolor_g << 8) | fgcolor_b; destLine[x] = 0xff000000 | (fgcolor_r << 16) | (fgcolor_g << 8) | fgcolor_b;
} }
zbufferLine[x] = posXW;
}
posXW += stepXW; posXW += stepXW;
posXU += stepXU; posXU += stepXU;
posXV += stepXV; posXV += stepXV;
@ -1474,15 +1481,11 @@ void ScreenTriangle::DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f); float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += light >> 7; // 255 -> 256 light += light >> 7; // 255 -> 256
float *zbufferLine = args->zbuffer + args->stencilPitch * 8 * y;
uint8_t *dest = (uint8_t*)args->dest; uint8_t *dest = (uint8_t*)args->dest;
uint8_t *destLine = dest + args->pitch * y; uint8_t *destLine = dest + args->pitch * y;
int x = x0; int x = x0;
while (x < x1) while (x < x1)
{
if (zbufferLine[x] <= posXW)
{ {
float rcpW = 0x01000000 / posXW; float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW); int32_t u = (int32_t)(posXU * rcpW);
@ -1501,7 +1504,6 @@ void ScreenTriangle::DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs
if (fgcolor != 0) if (fgcolor != 0)
destLine[x] = shadedfg; destLine[x] = shadedfg;
}
posXW += stepXW; posXW += stepXW;
posXU += stepXU; posXU += stepXU;