Mercurial > repos > blastem
changeset 710:4cd8823f79e3
First pass at emulating a vscroll latch. Titan's Overdrive demo seems to depend on the scroll value being latched early in the line before the HINT gets a chance to change it
| author | Michael Pavone <pavone@retrodev.com> |
|---|---|
| date | Sat, 16 May 2015 23:04:57 -0700 |
| parents | 3f065c20423c |
| children | 3342548de036 |
| files | vdp.c vdp.h |
| diffstat | 2 files changed, 7 insertions(+), 1 deletions(-) [+] |
line wrap: on
line diff
--- a/vdp.c Sat May 16 22:42:26 2015 -0700 +++ b/vdp.c Sat May 16 23:04:57 2015 -0700 @@ -657,7 +657,12 @@ vscroll <<= 1; vscroll |= 1; } - vscroll &= (context->vsram[(context->regs[REG_MODE_3] & BIT_VSCROLL ? (column-2)&63 : 0) + vsram_off] + line); + //TODO: Further research on vscroll latch behavior and the "first column bug" seen in Gynoug + //this should be close, but won't match the exact behavior Eke-Eke has written about + if (column == 2 || (column && (context->regs[REG_MODE_3] & BIT_VSCROLL))) { + context->vscroll_latch[vsram_off] = context->vsram[column - 2 + vsram_off]; + } + vscroll &= context->vscroll_latch[vsram_off] + line; context->v_offset = vscroll & v_offset_mask; //printf("%s | line %d, vsram: %d, vscroll: %d, v_offset: %d\n",(vsram_off ? "B" : "A"), line, context->vsram[context->regs[REG_MODE_3] & 0x4 ? column : 0], vscroll, context->v_offset); vscroll >>= vscroll_shift;
--- a/vdp.h Sat May 16 22:42:26 2015 -0700 +++ b/vdp.h Sat May 16 23:04:57 2015 -0700 @@ -143,6 +143,7 @@ uint32_t colors[CRAM_SIZE*3]; uint32_t debugcolors[1 << (3 + 1 + 1 + 1)];//3 bits for source, 1 bit for priority, 1 bit for shadow, 1 bit for hilight uint16_t vsram[VSRAM_SIZE]; + uint16_t vscroll_latch[2]; uint32_t frame; uint16_t vcounter; uint16_t hslot; //hcounter/2