Re: LD BC, nn

In article <1125669641.729139.232470@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
gareth <garethrichardadams@xxxxxxxxx> wrote:

[ Contention delays ]

>After reading the FAQ I came up with this:

Looks about right; some comments below.

>// Read op-code into variable opCode
>opCode = _spectrum.Memory.Peek(_PC);
>switch (opCode)
>{
> case 0x0: //00 NOP - This is the simplest Z80 instruction, and simply
>does nothing.
> _spectrum.TState += 4; //This instruction takes 4 clock cycles to

I'd suggest putting this outside the opcode switch as it happens for
every instruction not part of the opcode fetch.

> case 0x1: //01 LD BC,0x0000 - Load the next 16 bits into the BC
>register
> _spectrum.TState +=4; //The first part of the instruction (getting
>opCode) takes 4 clock cycles;
> _B = _spectrum.Memory.Peek((ushort)(_PC+1));

Should be _C, not _B.

> spectrum.TState +=3; //The second part of the instruction (loading

Again, I'd suggest making this part of Peek(); every memory access takes
at least three tstates.

Two more general hints:

1) Use autogenerated code. Writing out code for ~1000 separate opcodes
will just lead to errors.

2) Use a test suite to test your opcodes. There's one in the Fuse
distribution :-)

Cheers,

Phil

--
Philip Kendall <pak21@xxxxxxxxxxxxx>
http://www.srcf.ucam.org/~pak21/
.

Relevant Pages