1
0
mirror of https://github.com/TomHarte/CLK.git synced 2024-07-04 18:29:40 +00:00

I think this is correct. And have explained why.

This commit is contained in:
Thomas Harte 2016-08-20 16:54:43 -04:00
parent 7c9455251d
commit 8d839c5278

View File

@ -97,7 +97,12 @@ static uint8_t noise_pattern[] = {
#define shift(r) _shift_registers[r] = (_shift_registers[r] << 1) | (((_shift_registers[r]^0x80)&_control_registers[r]) >> 7)
#define increment(r) _shift_registers[r] = (_shift_registers[r]+1)%8191
#define update(r, m, up) _counters[r]++; if((_counters[r] >> m) == 0x7f) { up(r); _counters[r] = (unsigned int)(_control_registers[r]&0x7f) << m; }
#define update(r, m, up) _counters[r]++; if((_counters[r] >> m) == 0x80) { up(r); _counters[r] = (unsigned int)(_control_registers[r]&0x7f) << m; }
// Note on slightly askew test: as far as I can make out, if the value in the register is 0x7f then what's supposed to happen
// is that the 0x7f is loaded, on the next clocked cycle the Vic spots a 0x7f, pumps the output, reloads, etc. No increment
// ever occurs. It's conditional. I don't really want two conditionals if I can avoid it so I'm incrementing regardless and
// testing against 0x80. The effect should be the same: loading with 0x7f means an output update every cycle, loading with 0x7e
// means every second cycle, etc.
void Speaker::get_samples(unsigned int number_of_samples, int16_t *target)
{