i'll push and release the new code once i polish it and make it generic to sample rate and buffer sizes, and make it its own class so other emulator systems can use the same code whenever i get around to making a snes emulator or whatever

also as of right now the 404 pages may not display correctly on mobile devices or lower resolutions, basically just stuff getting cut off screen, or while maybe not completely off screen, positioned far down enough to look weird

the vblank flag is set on dot 1 of scanline 241, and cleared on dot 1 of scanline 261. i've never actually heard of any delays with nmi, or enabling nmi. i've only ever heard about when interrupts are polled. but implementing nmi like this fails the timing tests. i don't get it

everything i've read about how nmi works is that when the vblank flag is set, and nmi is enabled, the nmi line is pulled low, and the cpu uses a level detector to detect the falling edge of the nmi line. this is polled before the last cycle of most instructions, and then the cpu handles the nmi

like what do you mean you should be able to enable nmi at the end of the vblank and have it run, but also making that work breaks an earlier test. "it should happen but also it shouldn't" like what. am i missing something???

oh and i'm also nowhere near passing timing tests for irq or dmc dma either

or the seemingly contradictory delays which happen one way but not the other, or the test results from the same tests on different roms which contradict with each other

there are 3 stylized 404 pages which load randomly

it's up now!

as it turns out, making multiple 404 pages stylized based on games i like is really time consuming. i have two done so far (still not up)

i'm editing the website locally btw, the changes aren't up yet

sure, just throw them anywhere

and on top of that, i handle addressing separately from instructions, so even if i haven't added an unofficial instruction, the pc should still advance the correct number of bytes

the only thing i can think of that would stop the credits warp from working is if your emulator just gives up and crashes when running an unofficial instruction. mine defaults to behaving like a NOP for anything i haven't added yet, and i do have a good number of them added still

and here's the tutorial i followed: youtu.be/a4e6vLdvqlI although instead of stomping the second to last koopa in a specific place, i did the tail swipe shown in the above video because i found it easier

an explanation of how this works can be found here: youtu.be/fxZuzos7Auk

because dmc dma can happen mid-instruction, i basically accidentally created a real time corruptor for my emulator, lol. sometimes it would try to read a value, get interrupted by dma, and then happily store the now-unrelated value. this isn't the same as the hardware dpcm audio bug, but it's close