Issue with Megachip emulation of the LDHI instruction

[oshaboy]

The skip instructions should not skip the full LDHI instruction they only should skip one word even when before an LDHI instruction.

I can't attach the test rom I used but here it is in base64 YABjAmQOX/ABAF/wYAHwKdNFEhI=. The rom displays 0 on the original Megachip interpreter but 1 on Cadmium's Megachip mode.

[gulrak]

Hi and thank you for taking time to report this (and even with a test)!

TLDR: It is a deliberate choice.

The long form:

While the original MEGA-CHIP8 documentation doesn't specifically states it, and RS-M8001, the original emulator doesn't do it, we (in #chip-8 of the EmuDev Discord) decided to recommend it, when deciding on how to fill in the gaps of the documentation, as it is pretty underspecified. The result of that research was collected here. The wording in the SDK documentation is "Skip next instruction if..." so one could argue the modern interpretation is still following that wording. No program for the original RS-M8001 (we had access to at the time) is containing a situation where a skip is followed by a branch (at that would likely lead to terrible things), so we took the freedom.

If the argument is, that a program that follows that logic is run on RS-M8001 it would break, then I'd argue that Bxnn is broken in RS-M8001 too, or that it doesn't wrap the initial Dxyn coordinates while claiming , it actually has a bunch of bugs, doesn't follow the own documentation of MEGA-CHIP8 (it only supports 1MB ram while the documentation says it is 32MB, which is surprising as the ldhi only has enough bits for 16MB.

So to make a long story short, we "softly defined" something that is coined "modern MEGACHIP" and it implements the opcodes documented but ignored by RS-M8001 in a hopefully sensible way. Skipping full instructions is one of those decisions, having working blending is another one. It is partly inspired by Mega-8 by Ready4Next that also tried to fill some gaps and supports 16MB (but doesn't do long skips).

To make a program that runs on Revival Studio's implementation, Mega-8 and the recent emulators from people in #chip-8, one needs to stick to a safe subset, but if one wants to use blending the original is off the table anyway.

If you have a look at my Variant Detection Test you will see that it detects RS-M8001 as it's own variant because it is too broken to be seen as a bug-free implementation of MEGA-CHIP-8.