WIP: transpose software renderer

so far, level renders correctly
some patches also render correctly
everything else is still broken
This commit is contained in:
Alug committed 2025-10-10 20:18:19 +02:00
1 parent 513613d83b
commit f10470fadf
5 files changed
+75 -44

No files matched your search

+1 -1
View File
@@ -30,7 +30,7 @@ extern INT32 linesize;
FUNCINLINE static ATTRINLINE UINT8 *R_Address(INT32 px, INT32 py)
{
return renderscreen + (py + viewwindowy) * linesize + (viewwindowx + px);
return renderscreen + (py + viewwindowy) + (viewwindowx + px) * vid.height;
}
+2 -2
View File
@@ -187,7 +187,7 @@ static void R_DrawColumnTemplate(drawcolumndata_t *dc)
const intptr_t heightmask = dc->sourcelength-1;
constexpr INT32 npow2min = -1;
const INT32 npow2max = dc->sourcelength;
const INT32 stride = vid.width;
const INT32 stride = 1;
const UINT8 * restrict source = dc->source;
const lighttable_t * restrict colormap = dc->colormap;
@@ -343,7 +343,7 @@ void R_DrawFogColumn(drawcolumndata_t* dc)
// Framebuffer destination address.
dest = R_Address(dc->x, dc->yl);
const INT32 stride = vid.width;
const INT32 stride = 1;
const lighttable_t * restrict colormap = dc->colormap;
// Determine scaling, which is the only mapping to be done.
+26 -13
View File
@@ -107,9 +107,9 @@ static void R_DrawSpanTemplate(drawspandata_t* ds)
UINT8 * restrict dest = R_Address(ds->x1, ds->y);
UINT8 * restrict dsrc;
const UINT8 * restrict deststop = vid.screens[0] + vid.width * vid.height;
const UINT8 * restrict deststop = vid.screens[0] + vid.height *vid.width;
if (dest+8 > deststop)
if (dest > deststop)
{
return;
}
@@ -140,14 +140,14 @@ static void R_DrawSpanTemplate(drawspandata_t* ds)
if constexpr (Type & DS_RIPPLE)
{
dsrc = vid.screens[1] + (ds->y + ds->bgofs) * vid.width + ds->x1;
dsrc = vid.screens[1] + (ds->y + ds->bgofs) + ds->x1 * vid.height;
}
else
{
dsrc = dest;
}
while (count >= 8)
/*while (count >= 8)
{
// SoM: Why didn't I see this earlier? the spot variable is a waste now because we don't
// have the uber complicated math to calculate it now, so that was a memory write we didn't
@@ -167,7 +167,7 @@ static void R_DrawSpanTemplate(drawspandata_t* ds)
dsrc += 8;
count -= 8;
}
}*/
while (count-- && dest <= deststop)
{
@@ -175,8 +175,8 @@ static void R_DrawSpanTemplate(drawspandata_t* ds)
*dest = R_DrawSpanPixel<Type>(ds, dsrc, colormap, bit, source);
dest++;
dsrc++;
dest += vid.height;
dsrc += vid.height;
xposition += xstep;
yposition += ystep;
@@ -239,7 +239,7 @@ static void R_DrawTiltedSpanTemplate(drawspandata_t* ds)
const INT32 nflatxshift = ds->nflatxshift;
const INT32 nflatyshift = ds->nflatyshift;
const INT32 nflatmask = ds->nflatmask;
const INT32 stride = vid.width;
const INT32 stride = vid.height;
local_for_thread std::vector<INT32> tiltlighting;
iz = ds->szp.z + ds->szp.y*(centery-ds->y) + ds->szp.x*(ds->x1-centerx);
@@ -309,16 +309,29 @@ static void R_DrawTiltedSpanTemplate(drawspandata_t* ds)
x1 = ds->x1;
// not sure which one does a better job
/*for (i = SPANSIZE-1; i >= 0; i--)
{
bit = ((v >> nflatyshift) & nflatmask) | (u >> nflatxshift);
colormap = ds->planezlight[tiltlighting[ds->x1++]] + (ds->colormap - colormaps);
*dest = R_DrawSpanPixel<Type>(ds, dsrc, colormap, bit, source);
dest += stride;
dsrc += stride;
u += stepu;
v += stepv;
}*/
for (i = 0; i < SPANSIZE; i++)
{
bit = (((v + stepv * i) >> nflatyshift) & nflatmask) | ((u + stepu * i) >> nflatxshift);
colormap = ds->planezlight[tiltlighting[x1 + i]] + (ds->colormap - colormaps);
dest[i] = R_DrawSpanPixel<Type>(ds, &dsrc[i], colormap, bit, source);
dest[i * stride] = R_DrawSpanPixel<Type>(ds, &dsrc[i * stride], colormap, bit, source);
}
ds->x1 += SPANSIZE;
dest += SPANSIZE;
dsrc += SPANSIZE;
dest += SPANSIZE * stride;
dsrc += SPANSIZE * stride;
startu = endu;
startv = endv;
width -= SPANSIZE;
@@ -356,9 +369,9 @@ static void R_DrawTiltedSpanTemplate(drawspandata_t* ds)
bit = ((v >> ds->nflatyshift) & ds->nflatmask) | (u >> ds->nflatxshift);
colormap = ds->planezlight[tiltlighting[ds->x1]] + (ds->colormap - colormaps);
*dest = R_DrawSpanPixel<Type>(ds, dsrc, colormap, bit, source);
dest++;
dest += stride;
ds->x1++;
dsrc++;
dsrc += stride;
u += stepu;
v += stepv;
}
+16 -7
View File
@@ -1304,7 +1304,7 @@ static void SDLSetMode(INT32 width, INT32 height, SDL_bool fullscreen)
SDL_DestroyTexture(texture);
}
texture = SDL_CreateTexture(renderer, sw_texture_format, SDL_TEXTUREACCESS_STREAMING, width, height);
texture = SDL_CreateTexture(renderer, sw_texture_format, SDL_TEXTUREACCESS_STREAMING, height, width);
}
}
@@ -1339,6 +1339,7 @@ void I_UpdateNoBlit(void)
// I_FinishUpdate
//
static SDL_Rect src_rect = { 0, 0, 0, 0 };
static SDL_Rect dst_rect = { 0, 0, 0, 0 };
void I_FinishUpdate(void)
{
@@ -1368,13 +1369,13 @@ void I_FinishUpdate(void)
void *pixels;
int pitch;
SDL_LockTexture(texture, NULL, &pixels, &pitch);
int step = pitch / 4 - vid.width;
int step = pitch / 4 - vid.height;
UINT32 *restrict dst = pixels;
UINT8 *restrict src = vid.screens[0];
UINT32 *restrict palette = localPalette;
for (INT32 y = 0; y < vid.height; y++)
for (INT32 y = 0; y < vid.width; y++)
{
UINT8 *restrict end = src + vid.width;
UINT8 *restrict end = src + vid.height;
do *dst++ = palette[*src++];
while (src < end);
dst += step;
@@ -1382,7 +1383,8 @@ void I_FinishUpdate(void)
SDL_UnlockTexture(texture);
SDL_RenderClear(renderer);
SDL_RenderCopy(renderer, texture, &src_rect, NULL);
//SDL_RenderCopy(renderer, texture, &src_rect, NULL);
SDL_RenderCopyEx(renderer, texture, NULL, &dst_rect, 90.0, NULL, SDL_FLIP_VERTICAL);
SDL_RenderPresent(renderer);
}
#ifdef HWRENDER
@@ -1562,8 +1564,15 @@ INT32 VID_SetMode(INT32 modeNum)
SDLSetMode(vid.width, vid.height, USE_FULLSCREEN);
src_rect.w = vid.width;
src_rect.h = vid.height;
src_rect.w = vid.height;
src_rect.h = vid.width;
dst_rect.x = (vid.width - vid.height) / 2;
dst_rect.y = (vid.height - vid.width) / 2;
dst_rect.w = vid.height;
dst_rect.h = vid.width;
refresh_rate = VID_GetRefreshRate();
+30 -21
View File
@@ -614,9 +614,12 @@ static void CV_constextsize_OnChange(void)
// --------------------------------------------------------------------------
void VID_BlitLinearScreen(const UINT8 *restrict srcptr, UINT8 *restrict destptr, INT32 width, INT32 height, size_t srcrowbytes, size_t destrowbytes)
{
if (srcrowbytes == destrowbytes && srcrowbytes == (size_t)width)
//:skull:
return;
if (srcrowbytes == destrowbytes && srcrowbytes == (size_t)height)
{
size_t i = srcrowbytes * height;
size_t i = srcrowbytes * width;
#if defined(__SSE__)
while (i >= 16)
{
@@ -631,9 +634,9 @@ void VID_BlitLinearScreen(const UINT8 *restrict srcptr, UINT8 *restrict destptr,
}
else
{
while (height--)
while (width--)
{
M_Memcpy(destptr, srcptr, width);
M_Memcpy(destptr, srcptr, height);
destptr += destrowbytes;
srcptr += srcrowbytes;
@@ -791,13 +794,10 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
if (!desttop)
return;
deststop = desttop + vid.width * vid.height;
if (scrn & V_NOSCALESTART)
{
x >>= FRACBITS;
y >>= FRACBITS;
desttop += (y*vid.width) + x;
}
else
{
@@ -843,9 +843,9 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
y += (vid.height - (BASEVIDHEIGHT * dup)) / 2;
}
}
desttop += (y*vid.width) + x;
}
desttop += (x*vid.height) + y;
if (pscale != FRACUNIT) // scale width properly
{
@@ -858,11 +858,11 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
pwidth = patch->width * dup;
deststart = desttop;
destend = desttop + pwidth;
destend = desttop + pwidth*vid.height;
const INT32 stride = vid.width;
for (col = 0; (col>>FRACBITS) < patch->width; col += colfrac, ++offx, desttop++)
for (col = 0; (col>>FRACBITS) < patch->width; col += colfrac, ++offx, desttop += vid.height)
{
INT32 topdelta, prevdelta = -1;
if (scrn & V_FLIP) // offx is measured from right edge instead of left
@@ -894,13 +894,16 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
dest = desttop;
if (scrn & V_FLIP)
dest = deststart + (destend - desttop);
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup))*stride;
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup));
UINT8 *col_start = vid.screens[0] + ((x + offx)*vid.height);
deststop = col_start + vid.height;
for (ofs = 0; dest < deststop && (ofs>>FRACBITS) < column->length; ofs += rowfrac)
{
if (dest >= vid.screens[scrn&V_PARAMMASK]) // don't draw off the top of the screen (CRASH PREVENTION)
if (dest >= col_start) // don't draw off the top of the screen (CRASH PREVENTION)
*dest = source[ofs>>FRACBITS];
dest += stride;
dest++;
}
column = (const column_t *)((const UINT8 *)column + column->length + 4);
}
@@ -917,13 +920,16 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
dest = desttop;
if (scrn & V_FLIP)
dest = deststart + (destend - desttop);
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup))*stride;
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup));
UINT8 *col_start = vid.screens[0] + ((x + offx)*vid.height);
deststop = col_start + vid.height;
for (ofs = 0; dest < deststop && (ofs>>FRACBITS) < column->length; ofs += rowfrac)
{
if (dest >= vid.screens[scrn&V_PARAMMASK]) // don't draw off the top of the screen (CRASH PREVENTION)
if (dest >= col_start) // don't draw off the top of the screen (CRASH PREVENTION)
*dest = *(v_colormap + source[ofs>>FRACBITS]);
dest += stride;
dest++;
}
column = (const column_t *)((const UINT8 *)column + column->length + 4);
}
@@ -940,13 +946,16 @@ void V_DrawStretchyFixedPatch(fixed_t x, fixed_t y, fixed_t pscale, fixed_t vsca
dest = desttop;
if (scrn & V_FLIP)
dest = deststart + (destend - desttop);
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup))*stride;
dest += FixedInt(FixedMul(topdelta<<FRACBITS,vdup));
UINT8 *col_start = vid.screens[0] + ((x + offx)*vid.height);
deststop = col_start + vid.height;
for (ofs = 0; dest < deststop && (ofs>>FRACBITS) < column->length; ofs += rowfrac)
{
if (dest >= vid.screens[scrn&V_PARAMMASK]) // don't draw off the top of the screen (CRASH PREVENTION)
if (dest >= col_start) // don't draw off the top of the screen (CRASH PREVENTION)
*dest = *(v_translevel + ((source[ofs>>FRACBITS]<<8)&0xff00) + (*dest&0xff));
dest += stride;
dest++;
}
column = (const column_t *)((const UINT8 *)column + column->length + 4);
}
@@ -3810,7 +3819,7 @@ UINT8 GetColorLUTDirect(colorlookup_t *lut, UINT8 r, UINT8 g, UINT8 b)
void V_Init(void)
{
INT32 i;
INT32 screensize = vid.width * vid.height;
INT32 screensize = vid.height * vid.width;
for (i = 0; i < NUMSCREENS; i++)
{