voidderef
commented
3 years ago In GitLab by @icex2 on Oct 7, 2020, 20:55
Just a rough guess, but it might be that the game is polling IO not just every frame but, due to pumptool's piuio API layer, 4 times each frame. See the implementation of ptapi-io-piu-real.so in this line.
A quick hack to confirm that would be to just poll once, i.e. for a single sensor only, remember the current sensor cycle state and proceed with polling the next sensor on the next poll. Naturally, this reduces IO acuracy significantly but should re-create more accurate polling behavior regarding the CPU load.
Can you replace this entire function with the following, recompile, test and report back?
bool ptapi_io_piuio_recv(void)
{
static uint8_t i = 0;
/* cycle all four sensor groups */
//for (uint8_t i = 0; i < PTAPI_IO_PIUIO_SENSOR_GROUP_NUM; i++) {
piuio_drv_piuio_copy_outputs(i);
if (!piuio_drv_device_write(piuio_drv_piuio_out_buffer,
sizeof(piuio_drv_piuio_out_buffer))) {
return false;
}
if (!piuio_drv_device_read(piuio_drv_piuio_in_buffer,
sizeof(piuio_drv_piuio_in_buffer))) {
return false;
}
/* Invert pull ups */
for (uint8_t j = 0; j < 4; j++) {
piuio_drv_piuio_in_buffer[j] ^= 0xFF;
}
piuio_drv_piuio_copy_inputs(i);
//}
i++;
if (i >= PTAPI_IO_PIUIO_SENSOR_GROUP_NUM) {
i = 0;
}
return true;
}
super-triangle
In GitLab by @Cube on Oct 7, 2020, 20:39
Summary
When using the piuio hooks with pro and pro2, there is a significant bump in cpu usage. On mk9 hardware this is noticed by lag introduced during gameplay.
Expected behavior
Ideally, gameplay should not be affected on and no stuttering should be observed.
Current behavior
During gameplay, there is significant stuttering and dropped frames
Steps to reproduce
patch.piuio.emu_lib=./ptapi-io-piuio-real.soFurther things tested
Tested by disabling
patch.piuio.emu_libwhich lead to smooth gameplay (no input working of course)Context (Environment)
Pumptools version(s) affected
Master Branch
Game(s) and version(s) affected
Pro, and Pro2
OS version
pumpos
Hardware specs