mirror of
https://github.com/TomHarte/CLK.git
synced 2024-12-26 09:29:45 +00:00
Corrects delay for SN access.
This commit is contained in:
parent
56e691f256
commit
0dbd8a667d
@ -191,12 +191,13 @@ class ConcreteMachine:
|
||||
|
||||
// MARK: Z80::BusHandler
|
||||
forceinline HalfCycles perform_machine_cycle(const CPU::Z80::PartialMachineCycle &cycle) {
|
||||
// The SN76489 will use its ready line to trigger the Z80's wait for three
|
||||
// cycles when accessed. M1 cycles are extended by a single cycle. Short-circuit
|
||||
// that whole piece of communications by just accruing the time here if applicable.
|
||||
// The SN76489 will use its ready line to trigger the Z80's wait, which will add
|
||||
// thirty-one (!) cycles when accessed. M1 cycles are extended by a single cycle.
|
||||
// This code works out the delay up front in order to simplify execution flow, though
|
||||
// technically this is a little duplicative.
|
||||
HalfCycles penalty(0);
|
||||
if(cycle.operation == CPU::Z80::PartialMachineCycle::Output && ((*cycle.address >> 5) & 7) == 7) {
|
||||
penalty = HalfCycles(6);
|
||||
if((cycle.operation == CPU::Z80::PartialMachineCycle::Output || cycle.operation == CPU::Z80::PartialMachineCycle::Input) && ((*cycle.address >> 5) & 7) == 7) {
|
||||
penalty = HalfCycles(62);
|
||||
} else if(cycle.operation == CPU::Z80::PartialMachineCycle::ReadOpcode) {
|
||||
penalty = HalfCycles(2);
|
||||
}
|
||||
|
Loading…
Reference in New Issue
Block a user