Age | Commit message (Collapse) | Author |
|
Avoids float operations and redundant "1 -"
|
|
When DRC is on, game syncs to frame rate instead of audio buffer
capacity. Audio is resampled to generate more samples when buffer is
low and less when buffer is high, to keep buffer 40%-60% full. Buffer
size doubled to keep same avg. audio latency value.
Audio can distort when buffer is out of range, not often during
gameplay. Better resampler could improve but would be slower. When
buffer is always out of range (heavy frameskip), it is better to leave
off, DRC doesn't help anyway then.
Idea from RetroArch audio_driver.c and
https://near.sh/articles/audio/dynamic-rate-control
|
|
If core hasn't updated screen, no need to flip. Can save vsync delay,
makes ff faster
|
|
Small improvement to performance
|
|
This keeps from hanging when audio is sent before content is
loaded (pcsx_rearmed). picoarch cannot know correct frame rate /
sample rate before content is loaded (incorrect to call
retro_get_system_av_info then) so best to just skip it
|
|
|
|
If a core sets need_fullpath, content is patched and written to a file
in /tmp. This allows patching content for gpsp and probably others.
|
|
When starting without arguments, can select a core in the current
directory and some content (a game). Allows loading a different game
from the in-game menu.
|
|
|
|
|
|
|
|
|
|
|
|
Smooth is slower (10-15%) than a scaler built for a specific
resolution. Works well for downscaling and for odd screen ratios until
a custom scaler is built.
Replaces the snes smooth scaler (outperforms it) and the gba smooth
scaler (looks better)
|
|
|