Good improvement maybe to chop the dma, we don't want to run the dma for a very long time
and starve the frontend, another thing to do is to 


CYCLES_PER_FRAME = 1000
MAX_TO_RUN = 100

while self.cpu_frame < CYCLES_PER_FRAME {

    if self.excess = 0 {
        self.excess = self.cpu.run();
        self.cpu_frame += self.excess;
    }

    let to_run = min(self.excess, MAX_TO_RUN);
    self.excess -= to_run;
    self.clocks.run(to_run);
}

self.cpu_frame -= CYCLES_PER_FRAME;



loop 1:
    self.cpu_frame = 0
    self.excess = 0

    self.cpu_frame = 100
    self.excess = 100

    self.excess = 0 // done

    self.cpu_frame = 1100
    self.excess = 1000

    self.excess = 900

self.cpu_frame = 100

loop 2:
    self.cpu_frame = 100
    self.excess = 900

    self.excess = 800
    ...
    self.excess = 0

    self.cpu_frame = 200
    self.excess = 100

loop 11:
    self.cpu_frame = 0
    self.excess = 0

    self.cpu_frame = 10000
    self.excess = 10000

    self.excess = 9900

self.cpu_frame = 9000
self.cpu_frame = 8000
self.cpu_frame = 7000
... // doing 9 frame renders without clocking the compoenents, excess is still at 9900
self.cpu_frame = 0