Age | Commit message (Collapse) | Author |
|
|
|
context rather than trying to derive them from the cycle count. This should allow for more accurate handling of mid screen mode switches. Interrupt timing is broken currently though
|
|
vr debugger command. Fix horizontal interrupt bug. Slightly more accurate (but still broken) handling of switches between H32 and H40 modes.
|
|
and basic inspection of registers and memory work.
|
|
pushed. This commit represents the status of the working copy from that clone. It unfortunately contains some changes that I did not intend to commit yet, but this seems like the best option at the moment.
|
|
|
|
|
|
--HG--
branch : opengl
|
|
register flags
|
|
|
|
|
|
address 0 for reads to VSRAM at >= 40
|
|
DMA copy mostly correct from an observable ffect perspective. DMA copy probably does not reflect internal implementation still given that evidence seems to suggest no FIFO usage at all.
|
|
|
|
|
|
|
|
CRAM/VSRAM undefined bit results. Set number of bits actually saved in VSRAM to 11
|
|
|
|
|
|
and VSRAM bits can be implemented properly
|
|
stab at handling undefined bits of VSRAM and CRAM.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
visual observations of direct color DMA demos. Remove debug print statements.
|
|
|
|
completed if the FIFO is not empty
|
|
|
|
|
|
|
|
|
|
|
|
performance gain and to make it easier to use OpenGL for rendering
|
|
|
|
|
|
switching out of interlace mode.
|
|
|
|
|
|
|
|
with the cycle count getting out of sync with what I expect
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|