gb-test-roms VS mooneye-gb

Make sure you incorporate testing your implementation with test roms: https://github.com/retrio/gb-test-roms Do this as early in the project as possible. This will save you a huge migraine later on.

Have you tested your emulator against blargg's instruction timing test ROM? I believe this one only checks cycle counts, not memory timings: https://github.com/retrio/gb-test-roms/tree/master/instr_timing

Yeah the source for cpu_instrs is in github. https://github.com/retrio/gb-test-roms/tree/master/cpu_instrs/source

I actually have a question about this. I've been building a Gameboy emulator, the CPU works but when it runs the RET instruction the PC jumps to some really high address like 0x2000 or something when it's meant to jump to somewhere around 0x400 instead (after comparing it with another GB emulator). I think the RET address is meant to be read from HRAM, where the stack is stored, but when I look there it is the value around 0x2000 like I mentioned earlier. Are there any special quirks with this section of memory that I might not have implemented? I'm pretty much just reading and writing from an array for HRAM at the moment. I'm having the issue on one of blarggs cpu-instrs test roms https://github.com/retrio/gb-test-roms/blob/master/cpu_instrs/cpu_instrs.gb

So just start with implementing the CPU's fetch/decode/execute loop and a rudimentary memory bus, so that you can run simple test ROMs. You don't even need a display for many test ROMs (like Blargg's well-known test suite, which writes the results to the serial port as well as the display). After that you can do test-driven development (in a way) and focus on making more and more tests pass. Don't worry about details until you need to.

It looks like the test enables interrupts and then invokes the timer interrupt.

The Mooneye GB project, still only doing DMG and not supporting Color, is trying to accurately document as many variants and differences between versions as possible. But it's essentially at the cycle level and not gate level. If there are very nuanced differences at the cycle level, there are probably a lot more differences on the gate level.

Test your DAA instruction. https://github.com/Gekkio/mooneye-gb/blob/master/tests/acceptance/instr/daa.s

[3] https://github.com/Gekkio/mooneye-gb/tree/master/tests