bsnes/higan/sfc/ppu/sprite/sprite.cpp

220 lines
5.9 KiB
C++
Raw Normal View History

#include "list.cpp"
PPU::Sprite::Sprite(PPU& self) : self(self) {
}
auto PPU::Sprite::address_reset() -> void {
self.regs.oam_addr = self.regs.oam_baseaddr;
set_first_sprite();
}
auto PPU::Sprite::set_first_sprite() -> void {
regs.first_sprite = (self.regs.oam_priority == false ? 0 : (self.regs.oam_addr >> 2) & 127);
}
auto PPU::Sprite::frame() -> void {
regs.time_over = false;
regs.range_over = false;
}
auto PPU::Sprite::scanline() -> void {
t.x = 0;
t.y = self.vcounter();
t.item_count = 0;
t.tile_count = 0;
t.active = !t.active;
auto oam_item = t.item[t.active];
auto oam_tile = t.tile[t.active];
Update to v068r12 release. (there was no r11 release posted to the WIP thread) byuu says: This took ten hours of mind boggling insanity to pull off. It upgrades the S-PPU dot-based renderer to fetch one tile, and then output all of its pixels before fetching again. It sounds easy enough, but it's insanely difficult. I ended up taking one small shortcut, in that rather than fetch at -7, I fetch at the first instance where a tile is needed to plot to x=0. So if you have {-3 to +4 } as a tile, it fetches at -3. That won't work so well on hardware, if two BGs fetch at the same X offset, they won't have time. I have had no luck staggering the reads at BG1=-7, BG3=-5, etc. While I can shift and fetch just fine, what happens is that when a new tile is fetched in, that gives a new palette, priority, etc; and this ends up happening between two tiles which results in the right-most edges of the screen ending up with the wrong colors and such. Offset-per-tile is cheap as always. Although looking at it, I'm not sure how BG3 could pre-fetch, especially with the way one or two OPT modes can fetch two tiles. There's no magic in Hoffset caching yet, so the SMW1 pixel issue is still there. Mode 7 got a bugfix, it was off-by-one horizontally from the mosaic code. After re-designing the BG mosaic, I ended up needing a separate mosaic for Mode7, and in the process I fixed that bug. The obvious change is that the Chrono Trigger Mode7->Mode2 transition doesn't cause the pendulum to jump anymore. Windows were simplified just a tad. The range testing is shared for all modes now. Ironically, it's a bit slower, but I'll take less code over more speed for the accuracy core. Speaking of speed, because there's so much less calculations per pixel for BGs, performance for the entire emulator has gone up by 30% in the accuracy core. Pretty neat overall, I can maintain 60fps in all but, yeah you can guess can't you?
2010-09-04 03:36:03 +00:00
if(t.y == (!self.regs.overscan ? 225 : 240) && self.regs.display_disable == false) address_reset();
if(t.y >= (!self.regs.overscan ? 224 : 239)) return;
memset(oam_item, 0xff, 32); //default to invalid
for(unsigned i = 0; i < 34; i++) oam_tile[i].x = 0xffff; //default to invalid
for(unsigned i = 0; i < 128; i++) {
unsigned sprite = (regs.first_sprite + i) & 127;
if(on_scanline(list[sprite]) == false) continue;
if(t.item_count++ >= 32) break;
oam_item[t.item_count - 1] = sprite;
}
if(t.item_count > 0 && oam_item[t.item_count - 1] != 0xff) {
ppu.regs.oam_iaddr = 0x0200 + (oam_item[t.item_count - 1] >> 2);
}
}
auto PPU::Sprite::on_scanline(SpriteItem& sprite) -> bool {
if(sprite.x > 256 && (sprite.x + sprite.width() - 1) < 512) return false;
signed height = (regs.interlace == false ? sprite.height() : (sprite.height() >> 1));
if(t.y >= sprite.y && t.y < (sprite.y + height)) return true;
if((sprite.y + height) >= 256 && t.y < ((sprite.y + height) & 255)) return true;
return false;
}
auto PPU::Sprite::run() -> void {
output.main.priority = 0;
output.sub.priority = 0;
auto oam_tile = t.tile[!t.active];
unsigned priority_table[] = { regs.priority0, regs.priority1, regs.priority2, regs.priority3 };
unsigned x = t.x++;
for(unsigned n = 0; n < 34; n++) {
auto tile = oam_tile[n];
if(tile.x == 0xffff) break;
int px = x - sclip<9>(tile.x);
if(px & ~7) continue;
unsigned mask = 0x80 >> (tile.hflip == false ? px : 7 - px);
unsigned color;
color = ((bool)(tile.d0 & mask)) << 0;
color |= ((bool)(tile.d1 & mask)) << 1;
color |= ((bool)(tile.d2 & mask)) << 2;
color |= ((bool)(tile.d3 & mask)) << 3;
if(color) {
Update to v068r12 release. (there was no r11 release posted to the WIP thread) byuu says: This took ten hours of mind boggling insanity to pull off. It upgrades the S-PPU dot-based renderer to fetch one tile, and then output all of its pixels before fetching again. It sounds easy enough, but it's insanely difficult. I ended up taking one small shortcut, in that rather than fetch at -7, I fetch at the first instance where a tile is needed to plot to x=0. So if you have {-3 to +4 } as a tile, it fetches at -3. That won't work so well on hardware, if two BGs fetch at the same X offset, they won't have time. I have had no luck staggering the reads at BG1=-7, BG3=-5, etc. While I can shift and fetch just fine, what happens is that when a new tile is fetched in, that gives a new palette, priority, etc; and this ends up happening between two tiles which results in the right-most edges of the screen ending up with the wrong colors and such. Offset-per-tile is cheap as always. Although looking at it, I'm not sure how BG3 could pre-fetch, especially with the way one or two OPT modes can fetch two tiles. There's no magic in Hoffset caching yet, so the SMW1 pixel issue is still there. Mode 7 got a bugfix, it was off-by-one horizontally from the mosaic code. After re-designing the BG mosaic, I ended up needing a separate mosaic for Mode7, and in the process I fixed that bug. The obvious change is that the Chrono Trigger Mode7->Mode2 transition doesn't cause the pendulum to jump anymore. Windows were simplified just a tad. The range testing is shared for all modes now. Ironically, it's a bit slower, but I'll take less code over more speed for the accuracy core. Speaking of speed, because there's so much less calculations per pixel for BGs, performance for the entire emulator has gone up by 30% in the accuracy core. Pretty neat overall, I can maintain 60fps in all but, yeah you can guess can't you?
2010-09-04 03:36:03 +00:00
if(regs.main_enable) {
output.main.palette = tile.palette + color;
output.main.priority = priority_table[tile.priority];
}
Update to v068r12 release. (there was no r11 release posted to the WIP thread) byuu says: This took ten hours of mind boggling insanity to pull off. It upgrades the S-PPU dot-based renderer to fetch one tile, and then output all of its pixels before fetching again. It sounds easy enough, but it's insanely difficult. I ended up taking one small shortcut, in that rather than fetch at -7, I fetch at the first instance where a tile is needed to plot to x=0. So if you have {-3 to +4 } as a tile, it fetches at -3. That won't work so well on hardware, if two BGs fetch at the same X offset, they won't have time. I have had no luck staggering the reads at BG1=-7, BG3=-5, etc. While I can shift and fetch just fine, what happens is that when a new tile is fetched in, that gives a new palette, priority, etc; and this ends up happening between two tiles which results in the right-most edges of the screen ending up with the wrong colors and such. Offset-per-tile is cheap as always. Although looking at it, I'm not sure how BG3 could pre-fetch, especially with the way one or two OPT modes can fetch two tiles. There's no magic in Hoffset caching yet, so the SMW1 pixel issue is still there. Mode 7 got a bugfix, it was off-by-one horizontally from the mosaic code. After re-designing the BG mosaic, I ended up needing a separate mosaic for Mode7, and in the process I fixed that bug. The obvious change is that the Chrono Trigger Mode7->Mode2 transition doesn't cause the pendulum to jump anymore. Windows were simplified just a tad. The range testing is shared for all modes now. Ironically, it's a bit slower, but I'll take less code over more speed for the accuracy core. Speaking of speed, because there's so much less calculations per pixel for BGs, performance for the entire emulator has gone up by 30% in the accuracy core. Pretty neat overall, I can maintain 60fps in all but, yeah you can guess can't you?
2010-09-04 03:36:03 +00:00
if(regs.sub_enable) {
output.sub.palette = tile.palette + color;
output.sub.priority = priority_table[tile.priority];
}
}
}
}
auto PPU::Sprite::tilefetch() -> void {
auto oam_item = t.item[t.active];
auto oam_tile = t.tile[t.active];
for(signed i = 31; i >= 0; i--) {
if(oam_item[i] == 0xff) continue;
auto sprite = list[oam_item[i]];
unsigned tile_width = sprite.width() >> 3;
signed x = sprite.x;
signed y = (t.y - sprite.y) & 0xff;
if(regs.interlace) y <<= 1;
if(sprite.vflip) {
if(sprite.width() == sprite.height()) {
y = (sprite.height() - 1) - y;
} else if(y < sprite.width()) {
y = (sprite.width() - 1) - y;
} else {
y = sprite.width() + ((sprite.width() - 1) - (y - sprite.width()));
}
}
if(regs.interlace) {
y = (sprite.vflip == false ? y + self.field() : y - self.field());
}
x &= 511;
y &= 255;
uint16 tiledata_addr = regs.tiledata_addr;
uint16 chrx = (sprite.character >> 0) & 15;
uint16 chry = (sprite.character >> 4) & 15;
if(sprite.nameselect) {
tiledata_addr += (256 * 32) + (regs.nameselect << 13);
}
chry += (y >> 3);
chry &= 15;
chry <<= 4;
for(unsigned tx = 0; tx < tile_width; tx++) {
unsigned sx = (x + (tx << 3)) & 511;
if(x != 256 && sx >= 256 && (sx + 7) < 512) continue;
if(t.tile_count++ >= 34) break;
unsigned n = t.tile_count - 1;
oam_tile[n].x = sx;
oam_tile[n].priority = sprite.priority;
oam_tile[n].palette = 128 + (sprite.palette << 4);
oam_tile[n].hflip = sprite.hflip;
unsigned mx = (sprite.hflip == false) ? tx : ((tile_width - 1) - tx);
unsigned pos = tiledata_addr + ((chry + ((chrx + mx) & 15)) << 5);
uint16 addr = (pos & 0xffe0) + ((y & 7) * 2);
oam_tile[n].d0 = ppu.vram[addr + 0];
oam_tile[n].d1 = ppu.vram[addr + 1];
self.add_clocks(2);
oam_tile[n].d2 = ppu.vram[addr + 16];
oam_tile[n].d3 = ppu.vram[addr + 17];
self.add_clocks(2);
}
}
if(t.tile_count < 34) self.add_clocks((34 - t.tile_count) * 4);
regs.time_over |= (t.tile_count > 34);
regs.range_over |= (t.item_count > 32);
}
auto PPU::Sprite::reset() -> void {
for(unsigned i = 0; i < 128; i++) {
list[i].x = 0;
list[i].y = 0;
list[i].character = 0;
list[i].nameselect = 0;
list[i].vflip = 0;
list[i].hflip = 0;
list[i].priority = 0;
list[i].palette = 0;
list[i].size = 0;
}
synchronize();
t.x = 0;
t.y = 0;
t.item_count = 0;
t.tile_count = 0;
t.active = 0;
for(unsigned n = 0; n < 2; n++) {
memset(t.item[n], 0, 32);
for(unsigned i = 0; i < 34; i++) {
t.tile[n][i].x = 0;
t.tile[n][i].priority = 0;
t.tile[n][i].palette = 0;
t.tile[n][i].hflip = 0;
t.tile[n][i].d0 = 0;
t.tile[n][i].d1 = 0;
t.tile[n][i].d2 = 0;
t.tile[n][i].d3 = 0;
}
}
regs.main_enable = random(false);
regs.sub_enable = random(false);
regs.interlace = random(false);
regs.base_size = random(0);
regs.nameselect = random(0);
regs.tiledata_addr = (random(0x0000) & 3) << 14;
regs.first_sprite = 0;
regs.priority0 = 0;
regs.priority1 = 0;
regs.priority2 = 0;
regs.priority3 = 0;
regs.time_over = false;
regs.range_over = false;
output.main.palette = 0;
output.main.priority = 0;
output.sub.palette = 0;
output.sub.priority = 0;
}