The-Homebrew-Cloud/snes9xgx
2
0
Fork 0
mirror of https://github.com/dborth/snes9xgx.git synced 2024-11-27 13:04:21 +01:00
Code Issues Packages Projects Releases Wiki Activity

Speed optimizations for 2xBR, xBR-lv1 and DDT. (#1072)

Browse Source 
- Major speed ups. Now most games run at 60fps.
This commit is contained in:
Hyllian 2023-09-23 13:34:48 -03:00 committed by GitHub
parent 8b6ab62776
commit bb66556939
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 31 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

66
source/filter.cpp
View File

@ -667,14 +667,13 @@ static const uint16 lb_mask = LB_MASK565;
#define XBR(PE, PI, PH, PF, PG, PC, PD, PB, PA, N0, N1, N2, N3) \ #define XBR(PE, PI, PH, PF, PG, PC, PD, PB, PA, N0, N1, N2, N3) \
irlv1 = (PE!=PH && PE!=PF); \ if ( PE!=PH && PE!=PF )\
if ( irlv1 )\
{\ {\
wd1 = (df(PE,PC)+df(PE,PG))+(df(PH,PF)<<2); \ wd1 = df(PH,PF); \
wd2 = (df(PH,PD)+df(PF,PB))+(df(PE,PI)<<2); \ wd2 = df(PE,PI); \
if (wd1<wd2)\ if ((wd1<<1)<wd2)\
{\ {\
if ( ( !eq(PF,PB) && !eq(PF,PC) || !eq(PH,PD) && !eq(PH,PG) || eq(PE,PG) || eq(PE,PC)) )\ if ( !eq(PF,PB) && !eq(PF,PC) || !eq(PH,PD) && !eq(PH,PG) || eq(PE,PG) || eq(PE,PC) )\
{\ {\
dFG=df(PF,PG); dHC=df(PH,PC); \ dFG=df(PF,PG); dHC=df(PH,PC); \
irlv2u = (PE!=PC && PB!=PC); irlv2l = (PE!=PG && PD!=PG); px = (df(PE,PF) <= df(PE,PH)) ? PF : PH; \ irlv2u = (PE!=PC && PB!=PC); irlv2l = (PE!=PG && PD!=PG); px = (df(PE,PF) <= df(PE,PH)) ? PF : PH; \
@ -703,12 +702,13 @@ static const uint16 lb_mask = LB_MASK565;
}\ }\
}\ }\
#define XBRLV1(PE, PI, PH, PF, PG, PC, PD, PB, PA, N0, N1, N2, N3) \ #define XBRLV1(PE, PI, PH, PF, PG, PC, PD, PB, PA, N0, N1, N2, N3) \
irlv1 = (PE!=PH && PE!=PF); \ irlv1 = (PE!=PH && PE!=PF); \
if ( irlv1 )\ if ( irlv1 )\
{\ {\
wd1 = ((df(PE,PC)+df(PE,PG))+(df(PH,PF)<<1)); \ wd1 = df(PH,PF); \
wd2 = ( df(PH,PD)+df(PF,PB))+(df(PE,PI)<<1); \ wd2 = df(PE,PI); \
if (((wd1<<1)<wd2) && eq(PB,PD) && PB!=PF && PD!=PH)\ if (((wd1<<1)<wd2) && eq(PB,PD) && PB!=PF && PD!=PH)\
{\ {\
px = (df(PE,PF) <= df(PE,PH)) ? PF : PH; \ px = (df(PE,PF) <= df(PE,PH)) ? PF : PH; \
@ -722,7 +722,7 @@ static const uint16 lb_mask = LB_MASK565;
}\ }\
#define DDT(PE, PI, PH, PF, PG, PC, PD, PB, PA, N0, N1, N2, N3) \ #define DDT(PE, PI, PH, PF, N0, N1, N2, N3) \
wd1 = (df(PH,PF)); \ wd1 = (df(PH,PF)); \
wd2 = (df(PE,PI)); \ wd2 = (df(PE,PI)); \
if (wd1>wd2)\ if (wd1>wd2)\
@ -803,10 +803,13 @@ void Render2xBR (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
Ep[E0] = Ep[E1] = Ep[E2] = Ep[E3] = E; // 0, 1, 2, 3 Ep[E0] = Ep[E1] = Ep[E2] = Ep[E3] = E; // 0, 1, 2, 3
XBR( E, I, H, F, G, C, D, B, A, E0, E1, E2, E3); if ( (E!=F || E!=D) && (E!=H || E!=B) )
XBR( E, C, F, B, I, A, H, D, G, E2, E0, E3, E1); {
XBR( E, A, B, D, C, G, F, H, I, E3, E2, E1, E0); XBR( E, I, H, F, G, C, D, B, A, E0, E1, E2, E3);
XBR( E, G, D, H, A, I, B, F, C, E1, E3, E0, E2); XBR( E, C, F, B, I, A, H, D, G, E2, E0, E3, E1);
XBR( E, A, B, D, C, G, F, H, I, E3, E2, E1, E0);
XBR( E, G, D, H, A, I, B, F, C, E1, E3, E0, E2);
}
A= B; B=C; A= B; B=C;
D= E; E=F; D= E; E=F;
@ -828,8 +831,7 @@ void Render2xBRlv1 (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPit
} }
uint32 wd1, wd2; uint32 wd1, wd2;
uint32 irlv1, irlv2u, irlv2l; uint32 irlv1;
uint32 dFG, dHC;
uint32 E0, E1, E2, E3; uint32 E0, E1, E2, E3;
uint16 A, B, C, D, E, F, G, H, I, px; uint16 A, B, C, D, E, F, G, H, I, px;
@ -863,11 +865,14 @@ void Render2xBRlv1 (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPit
E3 = (i << 1) + nextlineDst + 1; E3 = (i << 1) + nextlineDst + 1;
Ep[E0] = Ep[E1] = Ep[E2] = Ep[E3] = E; // 0, 1, 2, 3 Ep[E0] = Ep[E1] = Ep[E2] = Ep[E3] = E; // 0, 1, 2, 3
XBRLV1( E, I, H, F, G, C, D, B, A, E0, E1, E2, E3); if ( (E!=F || E!=D) && (E!=H || E!=B) )
XBRLV1( E, C, F, B, I, A, H, D, G, E2, E0, E3, E1); {
XBRLV1( E, A, B, D, C, G, F, H, I, E3, E2, E1, E0); XBRLV1( E, I, H, F, G, C, D, B, A, E0, E1, E2, E3);
XBRLV1( E, G, D, H, A, I, B, F, C, E1, E3, E0, E2); XBRLV1( E, C, F, B, I, A, H, D, G, E2, E0, E3, E1);
XBRLV1( E, A, B, D, C, G, F, H, I, E3, E2, E1, E0);
XBRLV1( E, G, D, H, A, I, B, F, C, E1, E3, E0, E2);
}
A= B; B=C; A= B; B=C;
D= E; E=F; D= E; E=F;
@ -879,6 +884,7 @@ void Render2xBRlv1 (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPit
} }
} }
template<int GuiScale> template<int GuiScale>
void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height)
{ {
@ -889,11 +895,9 @@ void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
} }
uint32 wd1, wd2; uint32 wd1, wd2;
uint32 irlv1, irlv2u, irlv2l;
uint32 dFG, dHC;
uint32 E0, E1, E2, E3; uint32 E0, E1, E2, E3;
uint16 A, B, C, D, E, F, G, H, I, px, aux; uint16 E, F, H, I, aux;
uint32 nextlineSrc = srcPitch / sizeof(uint16); uint32 nextlineSrc = srcPitch / sizeof(uint16);
uint16 *p = (uint16 *)srcPtr; uint16 *p = (uint16 *)srcPtr;
@ -903,18 +907,10 @@ void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
while (height--) { while (height--) {
A = *(p - 1 - nextlineSrc);
B = *(p - nextlineSrc);
D = *(p - 1);
E = *(p); E = *(p);
G = *(p - 1 + nextlineSrc);
H = *(p + nextlineSrc); H = *(p + nextlineSrc);
for (int i = 0; i < width; i++) { for (int i = 0; i < width; i++) {
C = *(p + i + 1 - nextlineSrc);
F = *(p + i + 1); F = *(p + i + 1);
I = *(p + i + 1 + nextlineSrc); I = *(p + i + 1 + nextlineSrc);
@ -927,12 +923,11 @@ void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
if (E!=F || E!=H || F!=I || H!=I) if (E!=F || E!=H || F!=I || H!=I)
{ {
DDT( E, I, H, F, G, C, D, B, A, E0, E1, E2, E3); DDT( E, I, H, F, E0, E1, E2, E3);
} }
A= B; B=C; E=F;
D= E; E=F; H=I;
G= H; H=I;
} }
p += nextlineSrc; p += nextlineSrc;
@ -940,3 +935,4 @@ void RenderDDT (uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
} }
} }