#NES/FC/FDS/Dendy
1 messages Ā· Page 11 of 1
maybe try fresh, it's possible it captured some bad thing
alr
do need changed settings in advanced?
I feel like that Witch n Wiz bug report isn't happening. I played it today and it was fine. Issue 364
that's from a fresh game
Your clip sounds right to me
yeah save states restore the bad sound
because they captured the old behavior whatever it was
but it shouldnt happen in the future I guess
is this actually broken? https://github.com/MiSTer-devel/NES_MiSTer/issues/400
Nope, I play Pac-Man Championship Edition all the time on the core
Just played it yesterday
im gonna close that
great š
There's one dump of the game that is broken, has the wrong mapper or something, the proper one works
okay I think that is all the open game bugs addressed
minus mapper requests and the like
every test passes, no unexpected behavior from any demos
so, find more bugs for me please
Would it be possible to display the mapper somewhere once a game is loaded? I feel like that would be neat, but no idea what would be involved
Yeah, something in the menu so you could see if you were curious
benchmark said 126/128, is this necessary? š
if you turn on NES ppu behavior it will pass 15 column
19 should be passing, are you using the latest test core?
yea
what are your settings?
something must be impacting that
passes for me consistently
it's a pretty unimportant test all things considered, the 8-bit steins gate will show a single flickering pixel if it's working correctly
so am I correct in understanding that you should set the core to NES to pass all 128 tests?
famicom ppu behavior will fail test in page 15, but it's not so critical i think
it's actually better to have it fail
otherwise some games will break
it doesnt impact behavior
but yeah to pass the test you have to turn it on
I have no idea why the stale sprites is failing for you
It would be really useful to be able to tell what mapper a rom is from the core, that would be so cool. Especially if it also manages to work for ones not implemented.
yeah, I believe so
i would love to see those added. iirc it's a rather non-standard configuration, but those hacks are so cool, it would be a shame not to support them
VRC6 with CHR RAM should be fine? It might not work if CHR banks arenāt initialised properly
so the NES setting passes everything but the Famicom one doesn't? HAW HAW HOW THE MIGHTY HAVE FALLEN
yeah, I watched a video on them but then someone mentioned they weren't working in mister
It's official NES > Famicom
which Mega Man ROM hacks?
more like FamiCANT
the Rushjet1 music one for Rockman 2 works and that seems to use VRC6
MiSTer can load up oversized VRC6, just tried the MM2 ROM hack with RJ1's music, loads fine and CHR works too
huh really? last time i checked, it didn't work, so i'm glad to hear they do now. must've gotten fixed in an update
does the mm1 hack work as well?
i'm not at my mister right now but can check later if no one else can
MM1 doesn't
OK, I got it to boot. Header needed to be changed cause it was using an INES header
oh sweet!
It assumed there was no WRAM I believe
Edit the header in Mesen to NES 2.0 and change the options for WRAM and CHR-RAM to 8KB.
should look like this for MM1
I believe this applies for a lot of other mappers too, like I was able to get MMC5 + CHR-RAM working fine for a project
you have to force it to do outlandish things with headers, it wont be inaccurate by default, by design
okay ill give it til tomorrow and then merge the PR
I want to do a release after that so if you have something brewing @gray plaza get it in
No weird mappers from me haha
I really don't think there is any more mappers really that are worth adding now, plus I wouldn't want to make timings fail.
Core is now complete! Close channel! 
I pretty much found out about it by trying out the game on MiSTer and setting it to Vaus mode (already set up the NES core for mouse as spinner). Was not expecting it work (dpad control is actually pretty good). I get the full range on MiSTer, but not on real hardware. The only paddle that works with it is the Arkanoid II paddle, which can be finnicky with range (though has full range in Arkanoid II). It's also the only game I know of that, as far as I can tell, works with the Arkanoid II paddle, but not the black Arkanoid 1 paddle (which I didn't think was possible).
Awesome you discovered that. I'll post a video with examples of the iCode spinner and/or JogCon with Reflex Adapt controlling the game š
Thank you for the new build and explanation on Doreamon š
mapper 99 before release plz
Is anyone else having an issue where you save a game, it saves to the SD card, yet when you power the MiSTer on again, it wonāt recognize the save. Same with the settings. I see they saved to the microSD, but arenāt used. I have to constantly reconfigure the NES controller and despite me saving my game of Uncharted Waters, for a third time, it keeps telling me that I have no saved game despite the fact that itās on the MicroSD in the saves folder
Itās driving me crazy now
is this with the latest core in test builds?
Yep
Past three of them.
I have auto save on
Every time I need to backup the save, I press the menu button and let it save
lemme see if I can test it real quick
Okay
I just did a quick test with Zelda 2, which I had no saves in, and it saves just fine in the latest test core, showing up when I boot the game after rebooting the mister
Did you power it off and power it back on or just reboot?
Because mine was working on reboot too or game change, but as soon as I turned power off and came back later, it wasnāt working
first a reboot, just now I shut it off and turned it back on, save still exists.
I didn't touch anything that relates to saving
hmm
test saving with clear ram on
it's possible im clearing the saves?
no
it's working fine as far as I can tell
I just created a new save
Links to ROMs here: https://rainwarrior.ca/projects/nes/pico.html https://rainwarrior.ca/projects/nes/2a03puritans.html Download the NES ROMs and not the NSFs. The NSFs work fine. I tested on both ...
Does anyone have real hardware and a flashcart hooked up to test this out?
that's a mapper issue
Trying a couple other cores to see if they have the same issue
Makes sense. Is it an unofficial mapper?
Unless you're responding to someone else lmao
it's mapper 31
iNES Mapper 031 represents a mapper created to facilitate cartridge compilations of NSF music. It implements a common subset of the features used by NSFs.
PRG-ROM is bankswitched in 8 x 4 kB banks from $8000-FFFF. These are controlled by registers at $5FF8-$5FFF like the NSF mapper. The high bank at $F000-FFFF is initialized to the last bank at ...
I think there is a bug report for it
Trying to find a save point in Uncharted Waters
it's a stupidly simple mapper
Yeah. I'll close my issue as a dupe
So when loading an NSF, how does that work, because it seems to load it into a player ROM of some sort.
Press select when youāre in port
Super Metroid MSU-1 save worked fine so far. Powered off the MiSTer for a few minutes to be sure.
So at least I know another core is working as intended
Uncharted Waters has always been a picky game even with flash carts
Even if you try to do save states, you can fuck up the game. Doing the save states on the Everdrive N8 Pro to try and make money gambling is how I managed to see the ending. Plus, the game can error out if it detects too much āmemoryā being used. Itās really really strict.
I more use it because it was a game me and my stepdad would always play together but it comes in handy as a good test cart
this mapper is jank
I think it was added just for the NSF player and not tested with games
Do any games even use it? I thought it was just for playing NSFs
I can't get Unstable Waters' saves to work in the stable core either. So either there is an old bug there, or you can't save in the first city.
Both cores create a save
save states work fine in the new core
I never tried saving in the first city. Usually Iād do my first save after doing my first round to Bordeaux and back
i am not nearly good enough at this video game to go to france
You can tell I know the game like the back of my hand lol
Itās a screen up and two screens to the right
im pretty sure I found greenland
new core fixes a graphical issue that the stable has in the menu, so that's cool
Saves seem to work in the first city if you reset the core, but they are not detected ingame when loading a different game and loading uncharted waters again, or rebooting the mister, despite the file existing
Thatās the problem I had. While it was powered on, it recognized the save if I reset or changed games. After powering off and back on after a while, it wouldnāt
yeah, and it happens on the current (stable) core, so it's an old issue
Wonder why with this game
If itās happening on stable as well, maybe we can catch it this time
yeah, it's just good information so Kit knows it's not something she touched with this rewrite. it's been there
These past few NES betas are where Iām finally getting a chance to sit down and play it. Only had this MiSTer Pi about a month and mostly I been making NFC cards so havenāt had a chance to fully sit down sit down yet. Maybe a few minutes here or there. This time, I been sinking hours
Usually just sticking so far to games I had growing up that normally wouldnāt be played. Like Bigfoot, Days of Thunder, Dragonās Lair, Uncharted Waters, etc
My stepdad was sort of a racing and fantasy nut that also liked weird games (like Dino Riki and Caveman Games)
So when I played the NES, thatās mainly what we had
Time for a celebration š
RMW $2007 Extra Write still fails for me, but I'm using an old version of AccuracyCoin
Not that it passing makes any difference lmao
Latest version passes all tests
Also the PPU reset test passes in Famicom mode as well :P
I get a grey screen when trying to load Haratyler, a shmup homebrew. It seems to be a problem for everdrives as well?
Does there exist a fix or a workaround?
it means the mapper is not implemented
Ah gotcha
That proves that nobody is actually playing any games on their MiSTers! 
People are just trying abscure accuracy tests roms! 
I play fire emblem on the core. š
Itās a tiny toon adventures core for me
Did you mean a tiny goon adventure core?
For the people who donāt know me Iām just kidding. I say lots of silly bullshit like this as a parody of retro gaming culture.
I shitpost a lot too, so am I entitled to be a mod like you are, right @clever scarab ? 
I shouldnāt be a mod, this was a mistake lol
Adventures of Lolo core all the way (look at the JP releases, there is two disk system games as well as another cartridge game not released in the west called meikyuu no fukkatsu, very English friendly). Good shit if youāre into puzzle games
another day, another accuracycoin test to pass
Another day another build to pin. Not even sure if anyone even uses the pins but it makes me feel like Iām useful 
You ARE useful! You created pissfingers and introduced us all to gooning!
ā š©
š§æšš§æ
āØ
ā š«“
what do you need focused on with this test?
nothing really, I guess make sure safestates still work
basically the CPU only knew "reset" and "reset" isnt really the same as "power on"
so I had to add a "power on" thing for cold boots
it's really trivial
but no red squares allowed
cold boots happen when you load a rom
the signal already existed for APU behavior
happily, you dont have to enable nes ppu behavior anymore with the newest version of the test
Nooooooo!
Well something is wrong with that new build @fierce estuary
yep
Yeah the color of the sky is slighty off, right?
Lol
Oops
I forgot to remove experimental palette corruption
I'll upload a good build when I get home. Don't use that one
we have used at own risk
Please leave that as a toggle
Love it! 
kitrinx is making kaizo romhacks now?
Kaizo core
can speedrunners pass that?
The world as we know it falling apart
just tested benchmark on my AV Fami, and scores 125/128, page 10 fail first 3 steps
your AV famicon is not accurate - apologies
yea, true
same on my AV FC; those 3 tests failing is known behavior
here this is the same as the previous one sans palette experiments
Thank you Kitrinx - Super Mario is working great with the last rbf !
lol
I much preferred the original japanese version of super mario 1 from the last RBF. I hate that it's dumbed down for americans
Is the sky blue now?
I think it's better if you can really soak in the lovely periwinkle sky color with no distractions
Boiled down to its essence
and you got to watch it strip away all the layers, so you could understand all the cruft that you don't need
a real statement piece
actually it was deleting the palette colors one at a time
basically like the ending of hannibal where anthony hopkins is eating ray liota's brain, slice by slice?
oh sorry, spoilers
All say the movie is terrible. I oppose that.
I actually loved the book. I think they went the coward's way out with the movie
Let clarice and hannibal live happily ever after - who cares
they deserve happiness
they hook up in the book?
yeah. it's very romatic. he rescues her, drugs her, forces her to eat the dude's brains with him, and then they live happily ever after (not a joke - thats...uh...how the book ends)
I guess jodie foster declined to do the movie after reading that - and then they didn't use it.
an incredible cook
I like that book a lot. And, in the context of the whole thing, the ending makes more sense than I'm giving it credit for.
My dad and I listened to it in one sitting driving from Denver to Chicago, so there is likely a lot of nostalgia in there too
anyway, deleting the palettes is basically the same thing as marrying hannibal lector is what I'm trying to say
I understand completely
I'd marry a cannibal.
But what about the prion diseases
There are governments that advocate eating roadkill, I'd say we all don't have long so make the best out of it.
Does this new RBF fix the Uncharted Waters save issue or not yet?
This way I know if I should test it or not
I didnāt get the impression that that code got touched. Might be good to write an issue up for it
I thought that was the point of putting it here
Bug reports on discord can get lost in the ether and forgotten.
The best place would be to file an issue on the repo https://github.com/MiSTer-devel/NES_MiSTer
It was good to put it here to check
but yeah, once we verified that it was an issue, unless Kit (or whoever the core dev is) says they are looking at it, it's best to file it so it doesn't get lost
Like I was, near greenland
I'm so keeping this one as NES_CURSED.RBF
what's wrong with uncensored waters or whatever?
doesn't recognize saves after a reboot
will recognize them if you reset the core
but a full reboot results in "no saves found" even though the file is still there (and, assumedly, unchanged)
I'm naming my copy this too. I dream with the circuit bent version of the cores, moving the right analog stick or a mouse cursor to make the games glitch.
Though I was only able to get it to mess with Super Mario Bros
I just put it up
#1091056042667937944 message š¤
Someone pls send Kit some real hardware 
I need someone to run this rom on a real nes frontloader or newer, unmodified with nesrgb or anything
it depends on alignment, so you might have to reset it a few times to get the right alignment, but when you do, the colors should do funky stuff and not just scroll normally
I need a recording of that
maybe @hasty trout or @mint lichen
If nobody does it by tonight, I'll dig out my NES from the garage. I got an Everdrive N8 in it
people here have been really helpful so far
I can do it, gimme a few
thanks 
What rev PPU does the core attempt to replicate?
early model G
I noticed in SMB1, there isn't that half a scanline of flicker like on my real Rev G PPU
This core will become a 16bit core soon given the level of quality and accuracy 
I think something NMI related. There's a hack that fixes the behavior
Ah. So different PPUs behave differently or what? I get that behavior every time I play SMB1 on hardware
But I only have one NES
so, the CPU runs at masterclock/12 and the ppu runs at masterclock/4. When the system powers up, the alignments aren't always the same between the two
Oh. They don't use the same physical clock?
depending on which of the 4 alignments you get, it can have different outcomes
they both share a master clock, but the dividers are internal to each chip
so when they start counting differs
depending on just electrical randomness
Fair enough. Why would it start counting at different times, though? I thought all behavior in a pipeline stems from the clock
because the reset line can fall at slightly different times for analog reasons
Oh, interesting
or they just power up at different times for the same analog reasons
Is there a way to replicate that behavior in the core?
yes, and I plan to eventually, but it requires a whole big chunk of work
Also from what I read, Rev E PPUs don't have that issue. Any reason why, if that's true?
to really get it right the unstable bus when M2 rises and latching behavior of writes have to be properly replicated
I don't know much about revision E, but it's possible they finally synchronized them to only the "good" alignment electrically
one alignment is considered better than the rest, and most consoles get it more often than the others
For some reason, the Famicom seems to have Rev E PPUs as the most common one. I'll pick up a Famicom one of these days :P
oh E, they have alignment issues
famicoms and nes have some differences how they power up
Oh. Is it due to the CIC or smth?
no
they just tried to engineer it better with the NES because it came later
afaik nes never had any revision less than G
I thought super early ones had Rev E IIRC
Like, from the 1985 launch test in NYC
Not 100% sure, though
I suppose it's possible
Some page content & researching collaboratively taken from io55.net's NES page with permission.
There are two NES models, and three minor/internal revisions for both of the models. For the Famicom there are four main models, with several revisions between them.
Yeah. Early ones used Rev E. Just the vast majority of them used Rev G
Only reason I still use original hardware nowadays is for the native composite output. Hard to replicate
I can't get it to do anything other than scroll normally on my regular frontloader, reset a ton of times and it's the same scrolling. Sometimes the top left orange square half disappears but that's it.
Or was I supposed to manual scroll them and get this?
Just Down on Dpad til it got to the bottom
I think it's supposed to just happen on its own though
Immediately after a reset or after a while?
I'm not sure, that's why I need the recording š
well im probably going to not add this one, it's just cosmetic and arguably in an unappealing way
if they figure out how it works better I can add it
so where are we, any issues since the last test build? good to go?
thinking of merging all this if there's no further issues
I only tried a few games today but it seemed good. Tried a lot of homebrew yesterday and only noticed another accuracy improvement, haha.
k, time to ship it then
At this rate the core is going to have an option to simulate aging capacitors or clock resonators, oh wait, that last one is SNES
First game Iām gonna try is micro machines 
I'm going to play mario while humming the day the world went away
I canāt play Mario because heās stuck in an observatory in space right now
I can't play Donkey Kong because he's playing with rocks.
What is this meme?
Always thought DK was a crack head!
He becomes Cranky Kong when he doesn't get his rocks
Cracky Kong when it gets, maybe?
Is there a way to swap 12.5% and 25% pulse waves to emulate old famiclones?
is this hidden gem finally playable
never before
nope
why would you want to?
Nostalgia
it's a dangerous disease
The famiclone? I had a Super Joy when I was a kid lmao
The infamous one that there was a lawsuit over
Yeah lmao
What would changing the pulse waves actually do?
Why is PAL in black and white
color wasn't imported to europe until the 1990s
I said before, lots of people played many many hours with inverted duty cycle.
Too bad it's resources intensive to mess with it.
just tested the newest build and PAL myself, and yeah it's in blank and white
happens via HDMI or SCART
weird
@fierce estuary ^
damnit
the best laid plans of mice and men
it's super dumb
wire not_grayscale = ((in_draw_range || (vram_r_ppudata && is_pal_address)) && sys_type == 0) && ~grayscale_bit;
it's that line
typo that makes any system not ntsc grayscale
tee hee
Yeah I caught that earlier going through your code, was wondering when youād catch it.
testing me again, eh robby
Robby is so smart
Would it also be correct for the Dendy mode?
Given that many bootleg games were designed to be run on Famiclones, is it also possible several games were made with the inverted duty cycle in mind?
I kinda want faux stereo even though @fierce estuary said she didn't think it sounded good.
why don't add I just add an option for you Screech: No, Yes and you can just turn that on if you want your ears to bleed
Debug option, hotplate
no
For those dq3 speedrunners
watch me cook
don't I have console temperature as an option in 7800?
Haha
because atari stuff was so shitty, the temperature changed it's color output
Yeah for NES it's just like an option to give it a race condition I guess technically lol
And some people glitch by putting it on ice packs too
No, the Phantom System is PAL-M (a color system much closer to NTSC in resolution and refresh rate) and its duty cycle is reversed.
building a new thing now with no black and white
Can we get jail bars
you can
The moral of the story is that people should stop testing PAL
I've heard of light bulbs and TVs having warmer colors, but that's just ridiculous
yes I think so
It's an impressive port.
it might sound like bullshit but I think it's the best and most accessible version of Elite
seriously
The cat looks like he wants to force people to play that crappy Garfield game instead.
It's a reference to Bananza where DK befriends a talking rock
I tried to play it and could not even navigate the menus, trying to kludge those controls onto a controller with a d-pad and 4 buttons, only two of which are face buttons seems to have to taken it's toll
it's a bit cumbersome but I got used to it, the game starts with navigation computer already on your ship and that's a big factor in enjoying the game and actually landing the ship on stations
no one will know i'm cheating 
Ooh TAS support, thats awesome!
the problem is it's really hard to find verified scripts
so it ends up not being as good as tool for verifying regressions as I hope
kind of ostentacious
still cool to have
Oh that's neat! Is there a collected database/site of all the .tas files somewhere?
Guessing here is the best place
I didn't appreciate the format was used on so many different systems
Might be a good idea to try out TAS videos that do arbitrary code execution, especially some of those that do more than just use it to jump to the ending but use it just for the fun of showing off by reprogramming the game into flappy bird or something like that. I would imagine that's probably going to be really trying on the accuracy, depending on the game and exploit.
Hmm, seems files on that site are in all sorts of different formats
Ive got a BK2 file that might be worth testing, Iāll convert it to R08 later
oh, tas player feature on mister sounds nice
try this one out š
thank you for this improvment !
I used the test rom in my Twin Famicom and 5 tests are not passed. As I read this test is based in a model -G cpu and ppu. Anyway. I was doing some tests with mister one side, famicom the other side and they never desync. This is great.
a lot of those tas videos require very speciific memory configuration or some kind of weird frame timing that only emulators have
or arent verified on real hardware
So theyāre emulator speed runs?
tool assisted speedrun is what it stands for
Have you never watched the TAS block at a GDQ, Robby? Itās pretty wild what they do
oh my bad i mean theyāre strictly in the emulation camp right. Like you canāt do a TAS and assume itās possible on a real hardware if you could replicate it
Here's what I had on my mister, I haven't tested any of these recently
you can actually
they have a little device that plugs into the controller ports
it reads the polls
Ah wow
anyway that feature is not fully implemented
I think there's a list somewhere of ones that are Console verified, pretty sure there was more than this https://tasvideos.org/ConsoleVerification/Tests
Iām ok with that
As long as youāre okay with it, sorg
I think if thereās time and availability then including features for speed runners is nice but shouldnāt be a priority. They already have tools for that, just keep using that
This is not a tool for speedrunners
I've made one of those devices and played the same file back on emulator, on real nes, and mister though snac and the load tas feature Kitrinx was showing off.
thatās cool
@fierce estuary how are FDS movie files handled? Since I know the BIOS has a patch to load disk on button press
listen, when i spout hot takes I canāt be held back by silly things like āreading comprehensionā and āfactsā
Like a true American
Was this something you designed or an opensource project somewhere, by chance?
Is this flashing line in the boss select of Mega man 3 on real hardware too?
The line is flashing
Above shadow man
I'm pretty sure that's real hardware, but it's good to check
Played in real hardware. Played in a CRT TV
Longplay, Full game
=== Mega Man III / RockMan III ===
Mega Man 3 (stylized as Mega Man III) is a 1990 action-platform game developed and published by Capcom for the Nintendo Entertainment System. It is the third game of the original Mega Man series and was originally released in Japan on September 28...
Yeah sorry if this has been checked before
Perfectly corrupted on real hardware too šš»
I was going to say I remembered that from original hardware
Nothing escapes Kitrinx ā¤ļø
it's fine, that particular glitchy line hasn't been
When I had the cartridge back in Christmas 1990, I thought my copy was faulty.
But my friends had the same issue from another store.
I think I started with some open source project using an arduino with a sd card reader(it had a shift register like a nes controller and the nes controller plug), but eventually modified that to add buttons and a screen. The TASBOT used some other microcontroller
Glitched lines, dots and garbage. Getting this stuff nailed is incredibly impressive
will you keep the TAS support in the main release? I mean it can't hurt
Itās important because people grew up with emulators and have no concept of how the original hardware handled it
unless you're concerned about clueless people complaining about unverified runs
I love FPGAs pulling off these kinds of glitchy behavior because it gives things a feel of the real hardware
I want to add it as a feature, yes, but I need to give it more memory a couple options, and an on-screen indicator to keep the speedrunners from raging
ah yes, you mentioned about the indicator. In fact it could even be nice to keep it as an option for regular play for streamers
They aren't always strictly in the emulator camp, there are ones that are ran on a real system using an arduino as well.
Using an arduono sounds like CHEATING!
technically, in every sense, yes - https://youtu.be/PYX8iDqCDAY?si=M0Mi785oZGDGtKLw
the 2nd TAS they use a few minutes in is compatible with the mister core, i tested it this morning
YOU ARE ALL CHEATERS!!!!
It is normal.
oh sorry I meant COOL, silly autocorrect
These things happen too in the japanese version of SMB3 in the core. All these things are the true feeling.
the flickering and glitches should be fairly pixel perfect š
stuff like PAUSE flashing here in Isolated Warrior are pretty involved
As mentioned in the original issue I assumed it was a mistake with a missing check for submpper 15. That is the issue. It is a one line fix
do you want to do a quick PR? or I guess I can if you really dont want to
I was hoping to find why the nsf player is corrupted but am still looking at that.
tas stuff needs a setting for initial frames and filter types
so a lot will be broken til I add those
right now it just has a crude frame filter
also the initial ram for a lot is intended to be 00 00 00 00 FF FF FF FF pattern
you can do this by disabling ram clearing and using this rom for now
you run that rom to init the ram then run the game with the tas script already loaded
but ill add it natively
that rom is what they use on real consoles, apparently the SRAM can hold it's value fairly well for a couple of seconds after power off
don't worry about it too much though, I just wanted to test that PAL was okay now and that feature was already half implemented so I left it
Do FDS tas files work as well?
To be fair it can also be a useful tool for testing/debugging. If a TAS runs on real hardware but does not on the core you can try to figure out why.
Ok. Got a workaround for the nsfplayer
I still remember that one where they ran a TAS on two NES systems and a SNES system to execute code that basically turned the two NES systems into an audio input and SNES into a video input, then used them to play videos of TASes of other games like Mario64 or Portal by sending video data to the SNES and audio data to each NES through the controller ports.... wonder if you can replicate that with three misters 
For information, the save is working on Daikoukai Jidai (japanese version of Uncharted Waters)
After the whole reboot it doesn't work ... so same issue ...
it's very weird
i saw on MMC5 mapper page there is this line
"Uncharted Waters requires emulating bankswitching of PRG-RAM: it writes to PRG-RAM at one CPU address and expects to be able to read the same data back via a different CPU address."
aha, nice catch
wonder if that is it (he says, pretending he understands more than half the words in that sentence)
yes it is vague
Yeah as per usual my hot takes are wrong lol
I'm sure it is this
Can you add it on the GitHub issue?
Damn, you kick ass
@timber lava is pretty much like Winamp: he really whips the llamas ass! š¦
I thought that was implemented. I think Sim City used it as well.
Incidentally Sim City does a lot of unusual things with the ppu
Probably worth a regression test
It doesnāt work in the stable, but I havenāt gone back farther (further?) than that
ok ! done
Sim CIty uses also mapper MMC5?
yes
I did test with simcity throughout and it caught some regressions, but afaik it's working correctly as of now
Quick test looks ok to me. Just getting the title screen working the first time 6 years ago took days. The in game tool selections took a while too. Both look fine with latest.
and SimCity save also with a cold reboot we can use the save so maybe Uncharted Waters uses something differently
SimCity is very impressive on NES never tried it before
If uncharted Waters is writing to save RAM through banking outside of 0x6000-7fff then it won't trigger the need to save
assign save_written = (mapper_flags[7:0] == 8'h14) ? (prg_linaddr[21:18] == 4'b1111 && prg_write && prg_allow) : (prg_addr[15:13] == 3'b011 && prg_write) | bram_write;
Adding it to the mapper 20 exception on that line might fix it.
assign save_written = ((mapper_flags[7:0] == 8'h14) || (mapper_flags[7:0] == 8'h05)) ? (prg_linaddr[21:18] == 4'b1111 && prg_write && prg_allow) : (prg_addr[15:13] == 3'b011 && prg_write) | bram_write;
Made a build of the latest changes that adds the mapper 31 fix and nsf player fix by GreyRogue, and also the potential Uncharted Waters fix
Dang, you peeps have been pumpin on these NES updates. Every day I'm slamming my SD card in and out š
You can just use ftp
Hmm sadly I don't think this fixes Uncharted Waters, just tried it and my save still disappeared after rebooting the MiSTer
True š
Thank you MP2E for the build and GreyRogue for the proposition
I confirmed I have the same behavior for Daikoukai Jidai.
I tried also Load Backup Ram but still the same
Is it creating the save file? How big is it?
the save is created.
same size as SimCity
Uncharted Waters has two 8KB ram chips. Only one is battery backed. Maybe it is saving the wrong one https://nescartdb.com/profile/view/1052/uncharted-waters
Could try limiting to 16kb and/or inverting the bank bit https://github.com/MiSTer-devel/NES_MiSTer/blob/master/rtl/mappers/MMC5.sv#L601
assign prg_aout = {prgsel[7] ? {2'b00, prgsel[6:0]} : {6'b11_1100, prgsel[2:0]}, prg_ain[12:0]}; // 8kB banks
assign prg_aout = {prgsel[7] ? {2'b00, prgsel[6:0]} : {8'b11_1100_00, prgsel[0]}, prg_ain[12:0]};
or
assign prg_aout = {prgsel[7] ? {2'b00, prgsel[6:0]} : {6'b11_1100_00, !prgsel[0]}, prg_ain[12:0]};
If either of those work, a more generic change would be needed for other games.
@fierce estuary I'm super out of touch. Does this rework include the PPU/CPU syncing we discussed a while ago?
thanks! building new cores to test now
unfortunately no, but it's on my list
that's the one thing from my refactor that didn't make it in because of the sheer chance of error, so I wanted to break it up
Unfortunately, both prg_aout changes seem to have broken the game. When you take control of your character, character spawns out of bounds and has glitched graphics. When I try to save, it just hangs
just the way I remember it
your glitch is not as cool as my mario vanishing colors glitch
it's sort of cool and a little sad that now even real consoles with an everdrive or nesrgb aren't nessesarily accurate enough to use to check real hardware behavior to check the core
I agree! I had always just assumed that RGB mods were giving us exactly the right pixels out of the PPU, and now the MiSTer shows minor things that RGB mods canāt even do
this got all the way to the point where it tried to reset the console, but r08 format doesnt support that š
what's the size limit on the tas files?
i have an itsy bitsy 7,262 KB dragon warrior TAS file that isn't working for some reason and i'm pretty sure it's because it's 7,262 KB
How long does that take to play through?
like... 16 minutes i think if it's the same one
it might be doing resets but it's an r08 file so i'm confused
a boy and his blob gets very confused
New core works great for showing all the overscan area. Thank you!!!! šš¼
I noticed some games have a missing pixel at the top right. Has anyone noticed this as well? On real hardware? Here is a screenshot from Super Marioās Bros. 2
ninja gaiden is similarly very confused
probably uses sub-frame inputs as well
gradius is one of those very fun ones that because there is kind of a fixed time for most of each level they just go full wanker
What folder do the tas files live in?
i just put em in the games folder like roms in their own folder
i have the same set that greyrogue posted above, it's from the many years ago kitrinx first unveiled this
and then i added some from a github repo i found, not sure if they work
that's where i got the solar jetman one though
the gradius one is just insane lmao
Coincidentally was just going to enquire about this, with overscan set to everything noticed it on mario bros 3 and ninja gaiden 2 in the same place.
I rather like it
that looks like roughly scanline 0, pixel 245ish?
I can't see why anything would happen at that pixel at all
woah, it randomly went away when i reset the game
hrm
power on reset thingy maybe
yeah just started it, the like simultaneous left and right shots lol
Appears to be the pixel just before where the borders would be cropped with the "normal" overscan setting.
so the last pixel of the first scanline?
Last pixel I guess of what you would call the active area?
And yes it's on the first scanline.
I remember being very impressed with the metroid ones, think it was the first time I saw the door glitching. That's what you're talking about right?
a long time ago the metroid ones would fail, are they working now. aw you already answered š
good job on whatever you or someone else fixed š
nice, iirc samus would run into the spiky thing before the first door on the one I tried before. Like maybe it's starting position could have been determined by RNG using some value in ram, or could have just been a bad tas, not sure.
And running into the spiky thing messed the run up
@warm galleon Have you tried any TAS files that do code execution bugs to add or change things with the game? Curious if those work
No clue. Not very knowledgeable about the specifics of these files.
There might be some ACE (arbitrary code execution) runs in that lot you posted. The most well known are Mario 3, Mega Man 1, Final Fantasy, Kirby, could check those. Battletoads also has a 55 sec one that's looks to be in there, GEG (game end glitch)
Disabled vertical crop and set overscan to Vertical Overscan and itās really the last top right pixel that is missing
That corner shaving is as much of a CRT filter that Kitrinx is willing to implement.
And it caused her physical pain to do it.
She was like āalright, alright! Stop crying sheesh! There, are you happy now?ā š
im sure it probably has something to do with the uneven frames
actually it's not consistent
smb1 doesnt show it
crystalis also doesnt
maybe this is a thing a few games do?
Slightly messed it up. This looks like it works. https://github.com/MiSTer-devel/NES_MiSTer/pull/432/commits/e54719cf88237d75e4224a2ca9a737095926287c
Whoa, at the moment it's like you fix all the bugs that accumulated for years in a rad "coding spree" of just 2 weeks.
Would be nice if you could squeeze that in before having an official release.
a lot of the remaining ones are snac stuff that im not even sure is real, mapper requests, or just feature requests
Can you support that one mapper that uses a modern day processor to make crazy shit that wasnāt possible with the original NES
But how can we play Doom on the NES? š¤£
But my cycle accurate nostalgia
has nobody done this?
Kind of, but it involved using a Raspberry Pi as a co-processor, so I don't think that should count.
It was a joke, but someone kinda did.https://www.youtube.com/watch?v=FzVN9kIUNxw
The Raspberry did all the work, like a Super Game Boy.
And my reverse duty cycle nostalgia.
I was thnking more ones that don't just skip to the ending, but do things like give you new powers, or basically write an entirely different game in memory, or such. There is a SMB1 ACE TAS that after a few seconds of normal gameplay makes the NES play Bad Apple, IIRC it was made by the guy who made those recent NES accuracy tests, but I coulden't get it to work properly in Bizhawk even though it was made for it. Coulden't get it to work right in his own emulator either.
Isn't it just doom that does something like this?
I was talking about that one wacky RPG
What version of BizHawk were you using? Older versions fail to emulate Open Bus accurately, which that TAS uses
I don't remember what I was using, but I tried updating to the latest as of yesterday and... I keep getting LUA errors now when I try to run a TAS file
Talking about emulator behaviour accuracy. i guess mappers isn't included in the equation. they're not part of the console... well, it's a single circuit when a cartridge is connected, but you have to draw a line.
Is this for the script to convert to R08?
you're thinking too small
with the power of FPGA
Technically you have to draw a lot of lines, and transistors too
make a GPU for the Linux core that runs through the SNES
No, I was just loading a bizhawk format TAS by dragging and dropping, it was workiung in the previous version. Well, "working" in that it dind't crash, but the TAS file didn't play properly at a certai point
𤯠
Imagine attaching a RTX 5090 GPU to the MiSTer.
You can see every drop of sweat when you play CluClu Land
Ironically, you would be able to use it to heavily simulate a CRT over HDMI
Isn't there some new shader that emulates even the dot crawl or something like that, but needs an absurdly powerful GPU?
I read that as āstimulateā lol
No, that was just some very specific games
its not that much different than having a MIDI core to play with consoles' sound chips, except with PPU and other graphics chips š¤£
but you might as well make it a separate core at that point...
Sorg merged the mapper 31 fixes. Exciting
nosefart mapper?
wat lol
@fierce estuary do you do any work on this core? Sorg is doing all the updates it seems.
LOL OK for people who think Iām be g serious, itās just a joke because he does the nerves so it says his name
Merges
came out in 1998
Hopefully Sorg can add tas support, he has been on fire recently
That has the Uncharted Waters save fix too
nice!
@ember bone can you check uncharted waters with the latest unstable and see if it's fixed? if so, you can close the github ticket as well š
god that super mario bros 3 pixel is haunting me worse than those three assholes that come around and harass me about the future every christmas
Wait, all three are about the future?
I don't know I always tune them out
š¤£
everyone wants to argue about something. I just had someone tell me foam wasn't a food, and I insisted that bread was a foam, and was a food.
It's like people who argue what is and isn't a sandwich
I mean, tehnically bottled water is a sandwich
bread is a foam? wtflol
Does that make a hotdog a float?
Don't be telling me a tomato is a fruit next
how is bread not a foam?
it's air bubbles
styrofoam and similar things are considered foam
foams can be solids
bread is a foam
that makes sense lol
How can you bite down on something when your bottom jaw is actually biting upwards?
It is foamy, but saying it is a foam is going too far.
Thanks, I'm now looking up the scientific definition of what foam is
the videos posted earlier about Crystalis made me curious on the game. Have never played it, the snippets looked really cool
A CPU is just a rock that we tricked into thinking
Bread is a solid foam (colloid) according to me skimming the top results on google
It's a nice game, went quite far but didn't finished in the past, started playing again on Mister recently.
Gotta give that a try later today
So what's the deal with a missing pixel? That's quite odd. A bit of overscan would mask it but I'm intrigued now
I don't know if it's of any help but the pixel on this screen on Ninja Gaiden 2 at the start of the game doesn't stay solid but changes colour as the text comes onto screen.
I think it is an errant sprite pixel
Having a little walk about!
So many misbehaving pixels on the NES
OKAY so the reason that pixel doesn't draw is because in_visible_range_and_rendering only includes lines 0-239, not pre-rendering, and when it's disabled, the in_range flag is forced to false invalidating all loaded out sprites
pixel fixed, I think
Doesn't make sense to me yet. The pixel that's missing is on a line being displayed. Can you dumb it down a bit?
So the pixel is also missing from real hardware?
sprite evaluation basically tells the sprite shifters that all sprites evaluated on the pre-render scanline are bogus, and since sprites are loaded 1 scanline before they are shown, this means all sprite data on line 0 is ignored and invisible
Kit fixed it, so I assume not
every stray pixel is a rabbit hole
So the core is not accurate anymore? 
The core is perfect
Okay the pre render scan line (-1) is meant to be invisible so it uses the time (1/60 or 1/50) in the display cycle to do 'stuff' (shifter stuff). So there was a bug in that pre render time frame.
It was perfect already before hyper accuracy was implemented
That's all I got haha
Yep, but Kit has higher standards than perfection
Looks good on the games I and others mentioned, thank you!
š¶š¶š¶ that's why every cowbow sings the same sad song š¶š¶š¶
please give that some solid testing
there is this thing called Mousse that literally means foam. the chocolate one is great as dessert
With Pac-Man (USA) (Namco), I can now see part of the Pac-Man sprite in the far left corner.
If I remove the overscan part, we could see an orange line of 1px.
It's there on real hardware capture. https://www.youtube.com/watch?v=b6M-u1bBQjs https://www.youtube.com/shorts/ka8_b8lZvY0
This is just some of my gameplay without commentary from Pac-Man for the NES.
To see my full review of the game, go here: https://www.youtube.com/watch?v=cNhJoLw9UCY
Recorded from real hardware.
NSG: Videos for everyone who grew up in the 70ās, 80ās & 90ās.
Support the Show!
https://www.patreon.com/NoSwearGamer
Follow me on the social me...
Thanks for the confirmation.
Good catch!
Thatās interesting
Looks different here. Maybe thatās normal too.
Depends of the overscan option selected.
So far so good. Saved the game first time, turned the unit off for a few minutes and back on, save was still there. Saved a second time about a half hour later, left it for an hour while I cooked dinner, turned it back on after that, save will still working. So I think weāre good on that front. Right now, just came back from my first Santo Domingo trip of the game and all seems well.
I think thatās on legit hardware as well. I remember that.
almost no tv's except PVM's will see the top and bottom overscan
that's why 224 lines is considered "normal"
you're just seeing the bottom line of the wedge
I don't know how eye-rolly this is but Metroid's ENGAGE RIDLEY MOTHER F crashes the game consistently on recent core
I remember a while ago doing it and it went a bit beyond the dark screen
what is that?
Depends on which rom version you are using
Some glitchy password thing people figured out years ago.
This looks like a hardware capture to me. Seems like the correct behavior.
for the record, this password also crashes Nintendo's official emulators (I personally crashed both 3DS and Switch 1's NSO app)
okay, in the case of latter it was quitting to the NSO game library while politely stating "unknown error has occured"
No-Intro Headered
will try all of them later
I think I need some proof that that has different outcomes on a real system
since it relies on mapper behavior (this is an MMC1 related glitch) it has to be a real cart
I thought there were two revisions? I'll look. I thought there was a PRG0 and PRG1 in goodroms
m.r. only has listing for US, PAL and Virtual Console equivalents
I am wrong. There are known bad dumps of PRG0 which probably let you see more of the game after entering glitched password.
reference for various software emulators https://www.youtube.com/watch?v=WRrNmY5wZZI
Everyone heard about the ENGAGE RIDLEY MOTHER FUCKER password. Generally it is a password that makes the game crash, but it could have a different effect depending on the emulator you use. In some emulators it gives black screen, in others a glitchy version of the Brinstar area as well as in one of these it gives a glitchy version of Ridley's la...
I've never heard of Famtasia, this is new to me
it gets more wonderful later on
this is not worth being concerned about
yeah the current behavior is objectively correct
it hits a STP instruction
it should stop
I think it's correct behavior to crash
Dosen't it actually lock up the entire 3DS?
@fierce estuary I guess that your current TODO list is pretty empty now, the PPU rewrite/refactor was long in the planning. Mapper 99 support and TAS are possible targets in the future. But are there any tasks left, where an increase in accuracy could be made?
Kit sits and wait for a new failing accuracy test rom to appear 
I gave the NES core (updated unstable) a go last night...............WTF? looks so much better! 2A03Puritans one example. Used to look like a right jargled mess (graphics glitched to hell) Now when I put it on last night............cor! Can actually see it now
Well, that was a special problem of that game (more correctly: of the mapper implementation) and is not the result of the general accuracy gains.
to see these artifacts, neogeo transitions and pal artifacts, sega megadrive ram dots... all this is true feeling.
I want to implement variable cpu/ppu alignments
Why was that again? I believe there were good and no so good combo's
yes, there's very little value to the alternate alignments, except to speedrunners
what it means ?
for most people, slightly different glitchy pixels
for speedrunners, different RNG patterns and lag frames
@fierce estuary Its probably expected as I dont understand the pixel thing you fixed, but the corrupted pixels in Mega Man 3 boss select screen are still there. So still accurate
its probably more obvious in RNG behavior during gameplay
That's not something you can ever fix with any sort of emulation
Only if you decompile and port the game
I forgot what it was but the game was rushed with Capcom devs employing hacks and that line from the boundaries of the menu screen keeps bleeding into the stage select
And I don't think it was related to the other pixel thing
I didnt think either but thats why I commented on it
it only impacts line 0
If does differ for PAL NTSC. PAL has eight NTSC 4 possible CPU/PPU alignments. Disclaimer I'm not that good at math. So right now a fixed one is used?
PAL has no alignment
it constantly shifts alignments
it's horrible, and right now PAL is not really implemented great, but variable alignments would fix that
is it related to this "issue"?
https://github.com/MiSTer-devel/NES_MiSTer/issues/209
First I apologize if this does not contain enough information. I briefly mentioned that a friend spotted this on twitter and people jumped down my throat to file a bug report even though i have no ...
yes
I died twice in fire emblem today. Must be something wrong with the core, I'm an RPG Adonis and don't die k thx bye. 
I got the same number but blargg nesdev has a way better explanation ', there is one CPU clock every 16 master clocks. Therefore, VBL can begin at 16 different positions within a CPU clock. VBL occurs every 312341 PPU clocks. There are 5 master clocks in a PPU clock, so VBL occurs every 312341*5=531960 master clocks. This is a multiple of 8, but not 16, so VBL's position within a CPU clock will alternate between two points, each a half CPU clock apart. For a given power/reset, it will never fall on anything other than these two points in a CPU cycle. Since there are 16 points available, 16/2=8 possible states it could power/reset into'
But maybe there's an error somewhere
Got a request for a smb 2 JP skill wizard with an real Famicom. On world d-4 where the hammer bro is on the pipe. Can you wind assist glitch through the pipe by hitting back ššš ? I've seen it done but no dice for me.
I can't think of a reason it wouldn't work
Could llapi core be a reason?
I'll try the vanilla too but I'm frustrated now š
I have no idea if that's up to date or anything
is there any way to get the first line of pacman on real hardware capture?
like the very first line
I have this issue
there is some debate over whether that orange dot on the first line should be there or not on the real hardware
but it's very hard to get a real hardware capture with all the overscan
I see it now after messing with the scaling/cropping some more.
It seems to disappear and reappear as the game cycles through the attract screen
No problem
that orange dot actually is a rare situation, and it proves that sprites can be on line 0
the game loads 00 FF FF FF as the first sprite erroniously on line 239
the 00 is in range of scanline 0 as a y coordinate, and FF points off into memory space where an orange thing lives
FF as an X coord is dot 255
so you get an orange dot on line 0 at dot 255
This core is FFreakin accurate! 
Is there a way to check for the missing pixel on SMB3 on real hardware using the same approach used for the š ?
Or is it confirmed it was an issue on the core?
okay last one here, just make sure I didn't break anything horrible
so that orange dot is very special
it's a new discovery that fiskbit and I made
the other thing is this
this is testing the way that scanline corrupts there, and it's really hard to have it look perfect
it has to do with the behavior of disabling rendering mid-scanline
This core is awes00me!
im not afraid to say that it's the most accurate nes emulator there is, right now
Just caught up, amazing work š
I can't speak for every mapper but the cpu, apu, and ppu are really stellar
What have you done

anyone have any more bugs, pixels that aren't there that should be or vice versa, or other behavior that doesn't match the original?
hooray! you're awesome!
Ridiculous accomplishment
Woah, good job Kit! Real good job!
I will wait for 100th coin to update their video.
lol
The NES core and the MiSTer in general brings so much joy to my bleak life. Thank you so much for that.
But you are the star of your village
Not since the mister pi gave them running water. Now they're all about "hygiene" and have no time for stinky old me.
Ungrateful
Grats on such a rad core!! š„
it's really an amazing job. I know a lot of work went into it, but you made it look easy
Kitrinx fast running out of things to implement. Only a matter of time lads until there is nothing else for her to do and we can kill rabbits by blowing into an old PC microphone. Core endgame. š
Gotta add obscure modern mappers for weird games like that RPG that doesn't look like an NES game and has a stupid amount of animation. I forget what it's called. 
I'm joking of course. I would be surprised if that ever worked on the core.
š
Back to the Atari mines 
I never want people to question mega drive cram dot crawl ever again.
Who would think that this post would start a new era of perfection for the NES core? 
@slow mulch donāt you dare challenging Kit again! š
I was watching the video and then I ran to discord to see if anyone else had posted it š
Good luck finding new videos like that it seems
It was the spark that started this revolution and I still don't even know what a lot of this stuff means. Lol
Core is beyond perfection now! Thatās everything we need to know! 
I didn't even have the time to check the core yet, but reading about the development has been more fun than 99% of the NES library. Thanks Kitrinx!
Testing out the new core Kitrinx provided, can't seem to get R08 files to play back at all (probably something I am doing wrong), and I just noticed making a savestate or loading one changes my settings for overscan and ram initialisation
ill check this
im not getting that
also should ask if this input file works, i tried it with no success. used frame and polling (with and without cooldown) along with all possible offsets
tas stuff isnt really ready for prime time
ah gotcha, no worries
some of them dont work at all, some are too big (I think 260ish kb is the largest that will fit atm) some need specific versions of a game, some need to be toggled between "Polling" and "Polling w/Cooldown (frame)" and some need 0 or 1 blank frames
offset
however that one seems to work fine
method polling, offset 1
memory bizhawk
man, I ain't one to judge but Pepsi isn't a healthy beverage
Former Dawn, I actually do hope it would work someday. I think they are releasing the details of their mapper? It's not supposed to be the equivalent of some ARM cpu that the game just runs on
One of the devs of Sonic 3D blast, Sonic R, and a few other games had written homebrew back in the day to take advantage of the cram dots to display images with thousands of colors on the screen
weird, i can't seem to get it to work
What is limiting the size? Is it amount of free ram on the core/DE-10 Nano?
There was also a recent NES demoscene rom that called out Former Dawn as wasteful, since the whole point of the demo was that it didn't use a mapper XD
Surely that coould be streamed from hps after poc stage
why would it be wasteful? Genuine question
is that game actually out? it looked like the kind of thing that was going to just be in development for like 8 years til they lost interest and it fizzled out
I think it was just them having a playful jab at nes games that are impressive by using complicated mappers
Not yet, last big update they posted on their Kickstarter was a month ago so dosen't seem like it has fizzled out, at least not yet. Yeah, I meant if the game does actually come out someday then I would love if it would be possible to support it.
I wouldn't count on it
unless you think it's reasonable for a nes emulator to have to contain a modern ARM chip
more complex than the entire rest of the core
I thought the mapper was not just an ARM chip?
I have no idea, it keeps evolving but I thought it played mp3's and stuff
That was some other game I think
https://www.nesdev.org/wiki/NES_2.0_Mapper_451 <--- This one plays MP3s, and apparently there are two versions of that game, one that has MP3 support and one that does not
NES 2.0 Mapper 451 is used for the homebrew game Haratyler HP/MP. It is basically a homebrew TLROM-like circuit board that implements the MMC3 register's in an unusual fashion, and saves the high score to flash ROM. The game executes the AMIC A29040B flash ROM chip's "Software ID" command; if the manufacturer and model ID do not match the expect...
I hate fake mappers
I feel like if you can't make your game on vrc6 or mmc5 then you are fucking up
Seems like a big "feature" of that mapper 451 is just simply copy protection
Well, I can understand the need for mappers that support flash for new games since where are you going to get new stock MMC5s these days
on an fpga
Speaking of, they claimed that the NES RGB mods that rely on a FPGA or a CPLD(whatever that is) will apparently screw up displaying the game, but there are other RGB mods that will work.
Fpga? Never heard of it.
Just what is it doing to break RGB mods?
There was some more information about their mapper in the latest update but it didn't have much details
I like the sound of this game less every time it comes up
they should have just made this game for SNES instead
Now now, just because it can support the SNES controller and mouse does not mean...
the most recent trailer does make the game look and sound more like a snes title honestly
So why make games for old consoles of you're not planning to stay in those limitations and make it as compatible with the hardware/emulators as possible?
The whole thing is just a big contradiction. They want to say everything can be done on NES, but it's using a ridiculous custom mapper and is using a CDs worth of storage, so you are actually limiting who can play it on NES. They are fully making it on PC too, so why not just make a PC game? It's all really bizarre and uninteresting to me, as a huge NES homebrew follower, seems like they worked backwards from what they wanted. They keep announcing new features so I'd be surprised if it comes out within a couple years, or at all.
You could probably play crysis on nes of you put the whole pc in the cartridge. Guess the only limiting factor would be how much power the nes can supply to the cartridge slot? But then you could just be a sphincter and just make the "cartridge" use an extra outlet.
i remember a video from a few years ago of a guy who got doom "running" on nes, but it was just playing the game on an rpi and piping the video output through the nes
I feel exactly the same
Iād be pretty annoyed if I was a core dev and every time a shitty piece of homebrew came out, everyone asked why the core didnāt support it
Same thing happened with the MD and perineum
it is very annoying
people keep adding new crappy mappers rather than just using established ones and also trying to change existing ones to do dumb things
"lets just make mmc3 support 8 megabyte roms"
Developer equivalent of thinking youāre special
so now we're on mapper number 600 and something
the vast majority are minor revisions of existing mappers
most of them for games nobody will ever play
5 mappers covers something like 88% of the library
itās always xkcd 927
The US/EU version of estique uses an odd mapper to switch between the two MMC3 roms and doesn't even work correctly half the time.
Lemme guess⦠Standards?
Well, I can't comment on it myself, but the dev claims that the mapper is designed to have been technically feasible to produce around 1994 or so, it's not (at least they claim) basically just tossing a super powered ARM CPU on a cart and running the game off that.
Really, that's why I'm even showing any interest. If it turns out it really is just going the Doom route then I would lose all interest
Also for the record I have been trying to mess with the TAS files on here, hit or miss which ones work but I know the TAS feature is still just a test. https://github.com/alyosha-tas/NES_replay_files
or PC Engine...
im not really sure I have tas implemented correctly yet
so it might be desyncing because I implemented the feature wrong
I assumed that most would have problems since the feature is still unfinished, just wanted to see what would play. So far most either desync early or don't play at all.
Managed to get the SMB Max Score and Solar Jetman ones to play fully
several play fully for me
you can weed out any that are larger than about 250kb
they wont work
Am I using the feature wrong? Load the rom then load the r08 file right? Or are there other steps?
Sorry to swing this in the direction of Sega (again), but this is more an FPGA emulation question in general. The current Mega Drive core is a "wire by wire" replication of the original hardware which makes it super accurate (afaik). Based on my current understanding, the NES core isn't the same deal but it's also super accurate. I guess the question is how important is the wire by wire replication when it comes to accuracy?
Is it a "more than one way to skin a cat" deal?
There actually is a wire by wire core for Famicom now and I asked Kitrinx about it on GitHub back when it released.
This is her response(which I agree with now!):
https://github.com/MiSTer-devel/NES_MiSTer/issues/337#issuecomment-2263041410
Hello! I'm looking to adapt just the PPU code (ppu.sv and anything that it depends on) from this project to run on its own. It would run on a suitable FPGA chip with enough physical pins (>4...
That got pretty techy pretty quick but my understanding is that it can be accurate without needing to replicate all the unnecessary stuff?
Basically we would lose dejitter, ability to pause the core, save states, the simulation of ram decay, extra sprites and maybe more plus the code would be very difficult to work on
And at this point, I donāt know that we could even point to any tangible improvements
The current core is already absurdly accurate
Itās probably useful to simulate and compare but not so much as a direct port for users
Hmm, I wonder if that would cause issues for the Mega Drive core in the future
I.e. it fails accuracy tests
I do think that the old Genesis core shouldnāt have been shelved. The MegaDrive core is amazing but the Genesis one is much easier to work on and realistically, not that far from being perfect
I think it just needs the bus arbitration fixed
Yeah I noticed that it doesn't do the border thing in the Overdrive2 demo
I've also noticed some issues with CRAM dots not being accurate.
On the MegaDrive core?
Genesis core
Megadrive core is fine with the CRAM dots and border.
But back to NES, I've really been enjoying using the core in conjunction with Evercade playlists.
Good to know the core can run Bee 52
the difference is the old megadrive core did have a lot of accuracy problems, because the bus arbitration was not modeled well and nobody was acctively working on improving it
but I think the new code is so hard to work with and so hard to synthesize that basically development on the core is frozen because of this
the other problem with a wire-for-wire nes core would be that more than half the code in the core is it's mappers, not the system
Thanks for the explanation! I understand some of it š¤£
Kind of a shame about the old and new Megadrive cores tbh.
Not that the end user is going to notice an awful lot but it's kind of a shame that the new core is so complex that it's hard to work on.
My understanding is that most (if not all) mister core development is done for free, so I'd prefer that developers enjoyed working on the cores
it's not really complex, it's not human readable
it's like trying to edit a program after it's already been compiled
all those things are stuff the compiler would do under the hood if the code was written properly, but it's implementing them directly instead, which makes it just an unworkable pile of wires
How many of the cores are wire-by-wire/netlist based? Is it just Mega Drive and⦠Neo Geo?
And Jaguar before it got cleaned up?
I think so. a couple of chunks of atari 2600 are written that way but not much
jaguar
jaguar is still very much a netlist
Thanks for clarifying
sort of a special one though because it was a generated netlist
it was based on some old data format found on a discarded atari drive
generated into verilog, so they were able to add clock enables programattically
Wasnāt ao486 generated somehow too? I seem to remember someone mentioning some old Java code or something
(Java that did the generation)
Thatās really cool š
ao486 I think was somehow converted from the source code for Bocks
it's confusing to me
This has all been very interesting. I heard a lot of good stuff about the Mega Drive core being incredible due to it being made out of beheaded chips but it seems like there are a lot of downsides to that too.
i'd really be sad if someone tried to throw a netlist NES in there, because you'd lose a lot of cool stuff
And if the NES core can be this accurate without being wire for wire...
and it wouldnt really be better
The NES core is absolutely incredible
im really happy with it right now
Yeah the NES core is in an amazing state! I havenāt found any issues and Iāve tested a lot the past week or so
Do you play it much? š
There werenāt many issues before the recent changes anyway of course, itās been good for years
she beat accuracy coin
Canāt argue with that
The perfect core for perfect rpgs.
does it have speedup?
Dunno I never speed up.
I know there are people who want to speed everything up but for me it just ruins the atmosphere and the vibes.
RPGs are slow games that deserve to be savoured 
Just tried the Uncharted Waters save on the latest beta (10-10) and it loads again just fine. So this can be closed on that end.
i love the nes core as it is, and the improvements as of late only solidify that. i don't think we need a new ppu implementation, especially if it meant losing a lot of cool features we have now
i love the old genesis core too, and still prefer to use it over the newer mega drive one. i hope to see it come back into active development sometime
From my understanding it's just easy to do wire-by-wire since you're basically just kludging the original chip wiring into a FPGA, but it's not at all ideal because it's not made for a FPGA. You can still have just as accurate if not better with a core designed to run on a FPGA, you don't need wire-by-wire to do that. Also on top of all the other issues mentioned, it takes a large chunk of the FPGA to do. IIRC I recall it mentioned that it takes up nearly the entirety of the FPGA, pretty nuts when it takes up as much if not maybe even al ittle more of the FPGA than the PSX core. This also means that it's too big to add a 32X or CD to it so the SegaCD and 32X cores are still using the old Genesis core because of this.
Well, wire-by-wire is just the ultimate preservation goal. Being as close as possible. Maybe the Sega CD core can be made to play standard Mega Drive games as well? That could keep it being developed. Just indirectly.
Jotego had once a post that explained why netlists are a pain to work with. But even he ended up using itforbhis Sega System 18 core.
One the other hand that core is one that he is still battling with.