mirror of
https://github.com/libretro/parallel-dc.git
synced 2026-11-10 11:21:28 -10:00
retro_run asks GET_AUDIO_VIDEO_ENABLE before the frame is made, and with video left out a render to the screen is not drawn; the vblank reports no new picture. Renders to a texture are still drawn, the game reads them back, and so is a picture the processor wrote to the framebuffer, which is taken only once. In fast-forward the frontend's frameskip asks for a frame at the display's rate, so the GPU renders at that rate rather than at the emulation's: at a high internal resolution under threaded video the renders nobody saw queued up far enough to stall the picture. The headless test runs the disc with video asked for on one frame in four: the program's verdict and its sound are those of a run with every frame drawn, and far fewer frames are handed over.