- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of
registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
This commit is contained in:
Randy Heit
parent
14e94b86e2
commit
dda5ddd3c2
37 changed files with 1097 additions and 1276 deletions
178
src/r_draw.cpp
178
src/r_draw.cpp
|
|
@ -69,19 +69,6 @@ int scaledviewwidth;
|
|||
int viewwindowx;
|
||||
int viewwindowy;
|
||||
|
||||
extern "C" {
|
||||
int realviewwidth; // [RH] Physical width of view window
|
||||
int realviewheight; // [RH] Physical height of view window
|
||||
int detailxshift; // [RH] X shift for horizontal detail level
|
||||
int detailyshift; // [RH] Y shift for vertical detail level
|
||||
}
|
||||
|
||||
#ifdef USEASM
|
||||
extern "C" void STACK_ARGS DoubleHoriz_MMX (int height, int width, BYTE *dest, int pitch);
|
||||
extern "C" void STACK_ARGS DoubleHorizVert_MMX (int height, int width, BYTE *dest, int pitch);
|
||||
extern "C" void STACK_ARGS DoubleVert_ASM (int height, int width, BYTE *dest, int pitch);
|
||||
#endif
|
||||
|
||||
// [RH] Pointers to the different column drawers.
|
||||
// These get changed depending on the current
|
||||
// screen depth and asm/no asm.
|
||||
|
|
@ -130,8 +117,6 @@ const BYTE* bufplce[4];
|
|||
int dccount;
|
||||
}
|
||||
|
||||
cycle_t DetailDoubleCycles;
|
||||
|
||||
int dc_fillcolor;
|
||||
BYTE *dc_translation;
|
||||
BYTE shadetables[NUMCOLORMAPS*16*256];
|
||||
|
|
@ -161,7 +146,7 @@ EXTERN_CVAR (Int, r_columnmethod)
|
|||
/* */
|
||||
/************************************/
|
||||
|
||||
#ifndef USEASM
|
||||
#ifndef X86_ASM
|
||||
//
|
||||
// A column is a vertical slice/span from a wall texture that,
|
||||
// given the DOOM style restrictions on the view orientation,
|
||||
|
|
@ -212,7 +197,7 @@ void R_DrawColumnP_C (void)
|
|||
} while (--count);
|
||||
}
|
||||
}
|
||||
#endif // USEASM
|
||||
#endif
|
||||
|
||||
// [RH] Just fills a column with a color
|
||||
void R_FillColumnP (void)
|
||||
|
|
@ -404,7 +389,7 @@ void R_InitFuzzTable (int fuzzoff)
|
|||
}
|
||||
}
|
||||
|
||||
#ifndef USEASM
|
||||
#ifndef X86_ASM
|
||||
//
|
||||
// Creates a fuzzy image by copying pixels from adjacent ones above and below.
|
||||
// Used with an all black colormap, this could create the SHADOW effect,
|
||||
|
|
@ -480,7 +465,7 @@ void R_DrawFuzzColumnP_C (void)
|
|||
fuzzpos = fuzz;
|
||||
}
|
||||
}
|
||||
#endif // USEASM
|
||||
#endif
|
||||
|
||||
//
|
||||
// R_DrawTranlucentColumn
|
||||
|
|
@ -976,7 +961,7 @@ int dscount;
|
|||
|
||||
//
|
||||
// Draws the actual span.
|
||||
#if !defined(USEASM)
|
||||
#ifndef X86_ASM
|
||||
void R_DrawSpanP_C (void)
|
||||
{
|
||||
dsfixed_t xfrac;
|
||||
|
|
@ -1256,14 +1241,21 @@ void R_FillSpan (void)
|
|||
|
||||
// wallscan stuff, in C
|
||||
|
||||
#ifndef USEASM
|
||||
#ifndef X86_ASM
|
||||
static DWORD STACK_ARGS vlinec1 ();
|
||||
static void STACK_ARGS vlinec4 ();
|
||||
static int vlinebits;
|
||||
|
||||
DWORD (STACK_ARGS *dovline1)() = vlinec1;
|
||||
DWORD (STACK_ARGS *doprevline1)() = vlinec1;
|
||||
|
||||
#ifdef X64_ASM
|
||||
extern "C" static void vlinetallasm4();
|
||||
#define dovline4 vlinetallasm4
|
||||
extern "C" void setupvlinetallasm (int);
|
||||
#else
|
||||
static void STACK_ARGS vlinec4 ();
|
||||
void (STACK_ARGS *dovline4)() = vlinec4;
|
||||
#endif
|
||||
|
||||
static DWORD STACK_ARGS mvlinec1();
|
||||
static void STACK_ARGS mvlinec4();
|
||||
|
|
@ -1281,8 +1273,8 @@ DWORD STACK_ARGS prevlineasm1 ();
|
|||
DWORD STACK_ARGS vlinetallasm1 ();
|
||||
DWORD STACK_ARGS prevlinetallasm1 ();
|
||||
void STACK_ARGS vlineasm4 ();
|
||||
void STACK_ARGS vlinetallasm4 ();
|
||||
void STACK_ARGS vlinetallasmathlon4 ();
|
||||
void STACK_ARGS vlinetallasm4 ();
|
||||
void STACK_ARGS setupvlineasm (int);
|
||||
void STACK_ARGS setupvlinetallasm (int);
|
||||
|
||||
|
|
@ -1301,7 +1293,7 @@ void (STACK_ARGS *domvline4)() = mvlineasm4;
|
|||
|
||||
void setupvline (int fracbits)
|
||||
{
|
||||
#ifdef USEASM
|
||||
#ifdef X86_ASM
|
||||
if (CPU.Family <= 5)
|
||||
{
|
||||
if (fracbits >= 24)
|
||||
|
|
@ -1329,10 +1321,13 @@ void setupvline (int fracbits)
|
|||
}
|
||||
#else
|
||||
vlinebits = fracbits;
|
||||
#ifdef X64_ASM
|
||||
setupvlinetallasm(fracbits);
|
||||
#endif
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifndef USEASM
|
||||
#if !defined(X86_ASM)
|
||||
DWORD STACK_ARGS vlinec1 ()
|
||||
{
|
||||
DWORD fracstep = dc_iscale;
|
||||
|
|
@ -1374,7 +1369,7 @@ void STACK_ARGS vlinec4 ()
|
|||
|
||||
void setupmvline (int fracbits)
|
||||
{
|
||||
#if defined(USEASM)
|
||||
#if defined(X86_ASM)
|
||||
setupmvlineasm (fracbits);
|
||||
domvline1 = mvlineasm1;
|
||||
domvline4 = mvlineasm4;
|
||||
|
|
@ -1383,7 +1378,7 @@ void setupmvline (int fracbits)
|
|||
#endif
|
||||
}
|
||||
|
||||
#ifndef USEASM
|
||||
#if !defined(X86_ASM)
|
||||
DWORD STACK_ARGS mvlinec1 ()
|
||||
{
|
||||
DWORD fracstep = dc_iscale;
|
||||
|
|
@ -1863,17 +1858,17 @@ void R_DrawViewBorder (void)
|
|||
SB_state = screen->GetPageCount ();
|
||||
}
|
||||
|
||||
if (realviewwidth == SCREENWIDTH)
|
||||
if (viewwidth == SCREENWIDTH)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
R_DrawBorder (0, 0, SCREENWIDTH, viewwindowy);
|
||||
R_DrawBorder (0, viewwindowy, viewwindowx, realviewheight + viewwindowy);
|
||||
R_DrawBorder (viewwindowx + realviewwidth, viewwindowy, SCREENWIDTH, realviewheight + viewwindowy);
|
||||
R_DrawBorder (0, viewwindowy + realviewheight, SCREENWIDTH, ST_Y);
|
||||
R_DrawBorder (0, viewwindowy, viewwindowx, viewheight + viewwindowy);
|
||||
R_DrawBorder (viewwindowx + viewwidth, viewwindowy, SCREENWIDTH, viewheight + viewwindowy);
|
||||
R_DrawBorder (0, viewwindowy + viewheight, SCREENWIDTH, ST_Y);
|
||||
|
||||
M_DrawFrame (viewwindowx, viewwindowy, realviewwidth, realviewheight);
|
||||
M_DrawFrame (viewwindowx, viewwindowy, viewwidth, viewheight);
|
||||
V_MarkRect (0, 0, SCREENWIDTH, ST_Y);
|
||||
}
|
||||
|
||||
|
|
@ -1893,7 +1888,7 @@ void R_DrawTopBorder ()
|
|||
FTexture *p;
|
||||
int offset;
|
||||
|
||||
if (realviewwidth == SCREENWIDTH)
|
||||
if (viewwidth == SCREENWIDTH)
|
||||
return;
|
||||
|
||||
offset = gameinfo.border->offset;
|
||||
|
|
@ -1901,135 +1896,34 @@ void R_DrawTopBorder ()
|
|||
if (viewwindowy < 34)
|
||||
{
|
||||
R_DrawBorder (0, 0, viewwindowx, 34);
|
||||
R_DrawBorder (viewwindowx, 0, viewwindowx+realviewwidth, viewwindowy);
|
||||
R_DrawBorder (viewwindowx+realviewwidth, 0, SCREENWIDTH, 34);
|
||||
R_DrawBorder (viewwindowx, 0, viewwindowx + viewwidth, viewwindowy);
|
||||
R_DrawBorder (viewwindowx + viewwidth, 0, SCREENWIDTH, 34);
|
||||
p = TexMan(gameinfo.border->t);
|
||||
screen->FlatFill(viewwindowx, viewwindowy - p->GetHeight(),
|
||||
viewwindowx + realviewwidth, viewwindowy, p, true);
|
||||
viewwindowx + viewwidth, viewwindowy, p, true);
|
||||
|
||||
p = TexMan(gameinfo.border->l);
|
||||
screen->FlatFill(viewwindowx - p->GetWidth(), viewwindowy,
|
||||
viewwindowx, 35, p, true);
|
||||
p = TexMan(gameinfo.border->r);
|
||||
screen->FlatFill(viewwindowx + realviewwidth, viewwindowy,
|
||||
viewwindowx + realviewwidth + p->GetWidth(), 35, p, true);
|
||||
screen->FlatFill(viewwindowx + viewwidth, viewwindowy,
|
||||
viewwindowx + viewwidth + p->GetWidth(), 35, p, true);
|
||||
|
||||
p = TexMan(gameinfo.border->tl);
|
||||
screen->DrawTexture (p, viewwindowx-offset, viewwindowy - offset, TAG_DONE);
|
||||
screen->DrawTexture (p, viewwindowx - offset, viewwindowy - offset, TAG_DONE);
|
||||
|
||||
p = TexMan(gameinfo.border->tr);
|
||||
screen->DrawTexture (p, viewwindowx+realviewwidth, viewwindowy - offset, TAG_DONE);
|
||||
screen->DrawTexture (p, viewwindowx + viewwidth, viewwindowy - offset, TAG_DONE);
|
||||
}
|
||||
else
|
||||
{
|
||||
R_DrawBorder (0, 0, SCREENWIDTH, 34);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
// [RH] Double pixels in the view window horizontally
|
||||
// and/or vertically (or not at all).
|
||||
void R_DetailDouble ()
|
||||
{
|
||||
if (!viewactive) return;
|
||||
DetailDoubleCycles = 0;
|
||||
clock (DetailDoubleCycles);
|
||||
|
||||
switch ((detailxshift << 1) | detailyshift)
|
||||
{
|
||||
case 1: // y-double
|
||||
#ifdef USEASM
|
||||
DoubleVert_ASM (viewheight, viewwidth, dc_destorg, RenderTarget->GetPitch());
|
||||
#else
|
||||
{
|
||||
int rowsize = realviewwidth;
|
||||
int pitch = RenderTarget->GetPitch();
|
||||
int y;
|
||||
BYTE *line;
|
||||
|
||||
line = dc_destorg;
|
||||
for (y = viewheight; y != 0; --y, line += pitch<<1)
|
||||
{
|
||||
memcpy (line+pitch, line, rowsize);
|
||||
}
|
||||
}
|
||||
#endif
|
||||
break;
|
||||
|
||||
case 2: // x-double
|
||||
#ifdef USEASM
|
||||
if (CPU.bMMX && (viewwidth&15)==0)
|
||||
{
|
||||
DoubleHoriz_MMX (viewheight, viewwidth, dc_destorg+viewwidth, RenderTarget->GetPitch());
|
||||
}
|
||||
else
|
||||
#endif
|
||||
{
|
||||
int rowsize = viewwidth;
|
||||
int pitch = RenderTarget->GetPitch();
|
||||
int y,x;
|
||||
BYTE *linefrom, *lineto;
|
||||
|
||||
linefrom = dc_destorg;
|
||||
for (y = viewheight; y != 0; --y, linefrom += pitch)
|
||||
{
|
||||
lineto = linefrom - viewwidth;
|
||||
for (x = 0; x < rowsize; ++x)
|
||||
{
|
||||
BYTE c = linefrom[x];
|
||||
lineto[x*2] = c;
|
||||
lineto[x*2+1] = c;
|
||||
}
|
||||
}
|
||||
}
|
||||
break;
|
||||
|
||||
case 3: // x- and y-double
|
||||
#ifdef USEASM
|
||||
if (CPU.bMMX && (viewwidth&15)==0 && 0)
|
||||
{
|
||||
DoubleHorizVert_MMX (viewheight, viewwidth, dc_destorg+viewwidth, RenderTarget->GetPitch());
|
||||
}
|
||||
else
|
||||
#endif
|
||||
{
|
||||
int rowsize = viewwidth;
|
||||
int realpitch = RenderTarget->GetPitch();
|
||||
int pitch = realpitch << 1;
|
||||
int y,x;
|
||||
BYTE *linefrom, *lineto;
|
||||
|
||||
linefrom = dc_destorg;
|
||||
for (y = viewheight; y != 0; --y, linefrom += pitch)
|
||||
{
|
||||
lineto = linefrom - viewwidth;
|
||||
for (x = 0; x < rowsize; ++x)
|
||||
{
|
||||
BYTE c = linefrom[x];
|
||||
lineto[x*2] = c;
|
||||
lineto[x*2+1] = c;
|
||||
lineto[x*2+realpitch] = c;
|
||||
lineto[x*2+realpitch+1] = c;
|
||||
}
|
||||
}
|
||||
}
|
||||
break;
|
||||
}
|
||||
|
||||
unclock (DetailDoubleCycles);
|
||||
}
|
||||
|
||||
ADD_STAT(detail)
|
||||
{
|
||||
FString out;
|
||||
out.Format ("doubling = %04.1f ms", (double)DetailDoubleCycles * 1000 * SecondsPerCycle);
|
||||
return out;
|
||||
}
|
||||
|
||||
// [RH] Initialize the column drawer pointers
|
||||
void R_InitColumnDrawers ()
|
||||
{
|
||||
#ifdef USEASM
|
||||
#ifdef X86_ASM
|
||||
R_DrawColumn = R_DrawColumnP_ASM;
|
||||
R_DrawColumnHoriz = R_DrawColumnHorizP_ASM;
|
||||
R_DrawFuzzColumn = R_DrawFuzzColumnP_ASM;
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue