NES/FC/FDS/Dendy

[lost plume]

I think you need to run some jumper wires from the expansion port to the second controller port to use a Zapper as the Famicom gun used the 15 pin port, not the standard NES port

[granite pivot]

or make an NES adapter for the port. Misteraddons sold one.

https://misteraddons.com/collections/exclusive-products/products/nes-controllers-to-famicom-console-adapter

[granite pivot]

I have this and it definitely works with the Zapper. The Vaus for the US Arkanoid (ONLY US Arkanoid) also works using it. The US/JP Arkanoid controllers aren't compatible.

JP Arkanoid II Vaus isn't recommended for JP Arkanoid 1. It technically works, but doesn't give the full range.

[ripe shell]

dumb question

did OSSC slightly stretch the signal horizontally by default?

I know aspect ratios for NES and SNES are fickle subjective thing but can't help but shake this feeling

also, I can't unsee the crescent moon packing heat with a long barrel FAMAS

[fierce estuary]

looks that way

[ripe shell]

okay, MiSTer is definitely not outputting something similar to NESRGB signal, going by recommended settings the image is blurred out

[steep yew]

Can you set the 240p aspect to 4:3?

[ripe shell]

actually, yeah. there's 256x240 setting for Line*x outputs and with aspect correction and adjusting Video LPF to 16MHz it's pretty much pixel perfect now

but that's a little to narrow in my eyes but that's just me

okay yeah from cursory glance at SNES, MD, and PSX cores NES is like the only one that has odd horizontal stretch

[steep yew]

it's the way the scaler handles it. If it doesn't know the NES' PAR, it's going to just give you a W:H AR

Try loading up a street fighter arcade core. those are 384x224, so the OSSC might just give you a widescreen image

[ripe shell]

that's actually not bad

[steep yew]

sorry, I should have specified, any cps 2 street fighter

so super street fighter 2 and beyond

[fierce estuary]

could someone run this on a real frontloader nes and press left and tell me which of the numbers the triangles freak out on?

[latent breach]

5, 4,

[fierce estuary]

is that consistant?

[latent breach]

i get a double blinking image right below the main ones

[fierce estuary]

yeah that's the correct freak out

[latent breach]

yes considtent

[fierce estuary]

there's a bit of lack of consensus on which numbers it should do that on

[fierce estuary]

it's 5 and 4 not 3 and 5?

[latent breach]

5 and 4

using Everdrive N8 original if that matters

[fierce estuary]

k thank you

[latent breach]

no, thank you!

[hasty river]

https://tenor.com/view/no-thank-you-tracy-jordan-30rock-no-thanks-thanks-but-no-thanks-gif-21961959

[fierce estuary]

I had to make.. a few changes to the ppu to make those tests all pass

[fierce estuary]

run this through the ringer for me if you like

[hasty trout]

Will do!!

[timber lava]

Thank you for the new build 🙂
We get this time to time with the new build (like previous build)

[timber lava]

On clean boot, if we launch AccuracyCoins and pass all tests and load after 2nd2006 the screen is very unstable (screen shutters).

[hasty trout]

I tested all the games on the NESDev tricky-to-emulate games page and didn't find any issues, also tested games in closed issues on the NES core to make sure no old bugs resurfaced. Everything is looking really good

[clever scarab]

Is this core cycle accurate yet? I have standards and I don’t want to sully them.

Are we a Kitrinx or a Kan’t-rinx? Let’s get 127/127 passing!!!

@warm galleon I’m proud of that one

[main narwhal]

its 128 now...

[fierce estuary]

im not sure why you get those, but they seem stable for me

I think it has to do with memory state and uninitialized memory. he mentioned to me that he doesn't clear

[timber lava]

If it is only me maybe an issue with my MiSter ^^
I have tested plenty of games without issues 🙂

[fierce estuary]

I really need a high quality recording of micro machines recorded from composite on real hardware 🙁

there's one online but it's recorded at 30fps

so it misses important details

these white dots in the lower right of the character boxes is something no other emulator does afaik

[glacial turtle]

Might need to ask in the MDF/240p discords, sure one of those guys is set up for that if nobody here is

Unless one of the YouTube men can help. @split mauve maybe?

[hasty trout]

I can help

just setting up my twin famicom

[fierce estuary]

omg! can you get a recording from the title screen?

[hasty trout]

Sure! one moment

[fierce estuary]

the same as this: https://www.youtube.com/watch?v=AGxHeQ4kfUA

[fierce estuary]

pure gold

thank you so much

that game is like the gold standard for accurate timing of rendering enable, 2007 behavior, 2004 behavior, etc

[hasty trout]

No problem! Happy to help 👍

[dry lynx]

dot core wen

[half pasture]

NES-on-a-RBF wen

MiSTer wins again!

[fierce estuary]

the dot may seem trivial but it means that rendering_enabled is happening on exactly the right cycle and the palette ram is behaving absolutely perfectly. it has no tolerance at all for being off

there's actually no tests for some of this stuff so the isoteric behavior of weird games is the best way to tell

micro machines has never been emulated perfectly

[half pasture]

[hasty trout]

Outstanding! That’s really cool 😁

[timber lava]

even with NT Mini noir ?

[blissful prairie]

The last bastions of emulation resistance are slowly crumbling. First Metroid, now Micro Machines

[half pasture]

I think Analogue cores are far from being accurate

MiSTer does what Analoguedon’t

[dry lynx]

I was kidding but I really appreciate the effort you and others put into this stuff. It’s the getting the trivial stuff right the demonstrates how much attention to detail you/others have

[dry lynx]

Always has done

[split mauve]

What’s up?

Ahh I see. 60p capture. Yes I can but I just have a Twin Famicom

And not that game

[fierce estuary]

it's okay I got it

MP2E helped me out

[fierce estuary]

can you give this test another try? Turns out it didn't clear ram which caused issues getting an accurate output on real hardware

it should be either 3 and 5 or 4 and 6 and im not sure which of those real hardware does

it means 1 or 2 cycles of delay

[vivid roost]

always awesome to see a micromachines game running 100%

codemasters really pulled every trick in the book back then

[fierce estuary]

indeed

new accuracycoin dropped lol https://github.com/100thCoin/AccuracyCoin/tree/main

[half pasture]

They just discovered you’re going after each accuracy test and they wanna keep you busy 🤭

[graceful field]

The new hottest game on the nes

[dry lynx]

Xmas came early for Kit 😂

Better than more mappers in Chinese pirate carts

[dreamy marten]

I wonder if real hardware can pass these new tests

[dry lynx]

MiSTer > real hardware

[half pasture]

They can’t!

Unless they use a very specific PPU and CPU

MiSTer wins again!

[dreamy marten]

This guy has a weird hobby

[fierce estuary]

hmmm the 5/4 alignment

thanks

[mint lichen]

np, you are free to tag me or just send me any rom you need tested

[glacial turtle]

How many tests we up to now?

[half pasture]

Gotta pass’em all!!!

[latent breach]

I can test again shortly!

[fierce estuary]

it's okay I got one above

thanks though!

I think im all done

I have nothing left to fix

[glacial turtle]

So, I hear that people are asking for SNES mouse support on the NES core...

[gray plaza]

Wouldn't that work already just through using the SNES SNAC adapter?

[glacial turtle]

I don't think it does on the core, and usb mouse emulation isn't implemented

But all I know is half remembered GitHub issues

[gray plaza]

i sold my snes mouse a while ago so i couldn't tell you on that front

but i know snes pads work just fine on the nes core using the SNES SNAC adapter, along with the extra buttons (just need to code them into the games of choise because 8-bit registers are fun)

[half pasture]

NES core release wen

[latent breach]

I have a SNES mouse, but it doesn’t work with Reflex

been eyeing SNAX tho…

[urban meadow]

I appreciate the attempt at italics on the word "yet".

[steep yew]

meme generators can try to suppress my art

but I will prevail

[cursive sable]

Does anything currently out support that? I know that Former Dawn will, but that's not out yet

[glacial turtle]

I am partly winding up Kitrinx, but I did see requests for support on the NES GitHub

[cursive sable]

Actually curious about this. I noticed that 240pee thinks that X/Y/L/R are held down when I tried a SNES SNAC adapter on the NES core, no idea if that's what would happen on a real NES if used an adapter cable or not though, or if it's even a fault in 240pee itself

Also seems like there are apparently several homebrew NES games that have mouse support

[granite pivot]

There's a Maniac Mansion hack that supports mouse (technically more than one since there's an uncensored hack that incorporates it).

[fierce estuary]

I guess I should add the border

I promised it like 4 years ago

[solemn sedge]

Yeah I submittied a request. SNES mouse DOES work on NES core via SNAC with a SNES to NES adapter. I've used it to play the Hack of Maniac Mansion with mouse support. However, the SNES mouse is not exactly great. Would be awesome if we could use a USB mouse like we can in the SNES core. .......if you ever get bored @fierce estuary

[fierce estuary]

I'm prolly gonna add some new features

if games use it we can add it

[blissful prairie]

Border?

[fierce estuary]

yes nes had a border on the left and right

[clever scarab]

Unfortunately the Analogue NES consoles continue to have major bugs. Far from the best way to play NES these days.

[fierce estuary]

it doesn't show those white dots, and from the various tests, it's kind of mid on accuracy

[half pasture]

MiSTer wins again!

[timber lava]

Thank you for the informations 🙂

[clever scarab]

Although to be blunt, if I had an Analogue NES console I wouldn’t be able to tell.

I had an NES Mini and was perfectly happy lol

[timber lava]

I bought one NES Mini for one of my brother he is very happy also 🙂

[dry lynx]

You don’t get to decide that

The Internet decides what makes you happy and doesn’t

[clever scarab]

trolling @fierce estuary used to make me happy but now she ignores me

too bad for her I’m the 129th test

[fierce estuary]

but robby, 7 bits can only hold 128 values. are you the 8th bit?

if so, there's 127 tests beyond you

[dry lynx]

Robby’s always struck me as an integer overflow kinda guy

[clever scarab]

https://tenor.com/view/confused-math-what-wtf-peep-gif-6081931

no embed out of respect for dev channel but it’s the math lady gif lol

[half pasture]

Searching for “Nazare” works here lol

It’s the character name played by Renata Sorrah

Brazilian actress

[half pasture]

Sorry for the disrespect

[sullen fable]

If 100th Coin were here, I would've asked him if his accuracy test is possible on this particularly obscure Famicom accessory for the GBA: https://youtube.com/shorts/1bK2Rmjyiwc

[solemn sedge]

Ha, I just recently imported one of those. It's terrible and great at the same time.

[sullen fable]

Great because you get to play NES games on the go, but horrible to play?

[solemn sedge]

Great because it has the audacity and absurdity to exist. Horrible because it's janky, finicky, uncomfortable and poor video quality. Though, all that said, I prefer it to the NES emulator releases they did.

[gray plaza]

It’s just a NOAC so it’d likely fail the exact same tests as other epoxy blob NOAC’s

[timber lava]

New accurcyCoin update 👍

[gray plaza]

This one replaces RMW $2007 right?

[timber lava]

Yes there is 4 commits and one mention $2007

[gray plaza]

Ok, 100th coin did mention in a VC he was going to remove it cause of it being console dependent

[fierce estuary]

he added a new test for a new quirk I figured out with fiskbit

heavy lies the head...

[timber lava]

Great milestone 🙂

[steep yew]

Killing it, Kit

[covert crypt]

So it's now more accurate than an NES?

Also, I just want to say I love Kitrinx's scanlines settings. It's refreshing to see someone not hammering the scanlines.

[half pasture]

MiSTer NES core is officially the BEST NES EVAH!!!

[fierce estuary]

Better Than The Real Thing ™

[half pasture]

Kitrinx does what Nintendon’t

[fierce estuary]

it's probably more accurate than a real nes with an everdrive at least

[half pasture]

Time to throw our Analogue NES consoles away

[gray plaza]

Would it be more accurate than a NES with the RGB mod? Cause I'm sure I had some tests fail with my NESRGB AV Fami

[fierce estuary]

I accidentally found and fixed a bug in MMC1 too

[fierce estuary]

you're gonna have a hard time passing all those tests with real hardware

pff

that janky old stuff can barely run

[gray plaza]

Ok yea it was the PPU read buffer test ROM that had issues

[blissful prairie]

I guess that having Metroid and Micromachines running perfectly now is the better outcome of those tests than showing off xxx/128.

[half pasture]

Debatable 😌

[livid leaf]

It's always worth it for Micro Machines

[glacial turtle]

Amazing work. I hope we get some update videos detailing that MiSTer now passes all the tests

[steep yew]

https://tenor.com/e3NatgkPrB1.gif

[simple hedge]

Would be cool if there was a way to connect a cart reader to MiSTer so we could play modern homebrews and aftermarket games.

[cursive sable]

Oh right, this test is mapperless right? I guess it would not really be possible to make a test for the different mappers since you would need a different physical cart for each with a harvested mapper from a donor cart.

Not like you can just buy one from 30yearoldnesmappersforsale.com

... that should be 40 year old

[cursive sable]

A cart reader like Sanni's is generally for dumping carts, you would just need a cartridge port and have it connected to the FPGA, thing is, I don't believe there are enough free pins on the Mister to do that even if you gave up analog

Though there are some tech demos of that being done with gameboy carts on the superstation 2

Would be cool to do though, would not have to worry about support for mappers/special chips in that scenario since it would be running off the cart and using it's hardware

Would just have to have the base console hardware emulated accurately on the fpga

[fierce estuary]

right

[glacial turtle]

If there are any worthwhile games with mappers the core doesn't support then let us know and maybe someone can add support.

[gray plaza]

Oh boy testing for MMC5A would be so fun!

[clever scarab]

@fierce estuary what was that term I was trolling about earlier about this core not passing, started with a d

I forgot it

[fierce estuary]

nothing about MMC5 is fun

[clever scarab]

DMA!!!!

What about the DMA

[blissful prairie]

Let's hope that Furrtek will finish his MMC5 Reverse Engineering. With that in place a highly accurate mapper implementation is at least possible.

[clever scarab]

Here it is - https://www.nesdev.org/wiki/MMC5

[fierce estuary]

it's already highly accurate

[blissful prairie]

I wonder why that was so high on his agenda. Is there something still unknown about it?

[clever scarab]

Could’ve been for fun

[blissful prairie]

🫠🫠

[gray plaza]

Yep, it’s got so many features that I imagine were a nightmare to implement

And also there’s another variant of the mapper with even more features that no games ever used

[cursive sable]

For me personally the only mappers I would like to see that there isn't support for yet are the homebrew flashrom mappers, there are still some that are not supported, also the ones that are don't support saving to the flashrom. From what I understand though they are also quite complex.

[cursive sable]

The devs who made Lemmings, Body Harvest, and the first three GTA games? What about them, they only ever made a single NES game.

😉

[cursive sable]

The link Robby posted says that there are quite a few functions that are unknown, and some more unknowns with the MMC5A though nothing ever used it's additional functions

[jade zenith]

just breed 🥵

[glacial turtle]

Was having a look through the smorgasbord of issues on GitHub. I suspect this one may be fixed now if anyone is testing:
https://github.com/MiSTer-devel/NES_MiSTer/issues/192

My favourite issue so far is renaming "periphery" to "peripheral" in the input menu. Which, as pedantic as that is, is fair. That would make more sense.

[grand jewel]

Wark tested it some days ago.
#1091056042667937944 message

Same for Eliminator Boat Duel.

[glacial turtle]

Oh nice, missed those posts. Probably worth commenting on the issues that are fixed so they can be closed off

[dreamy marten]

Are you suggesting that he will make more tests designed for mister to fail?

[mellow dawn]

I did at one point buy a copy of Bandit Kings of Ancient China to see if it used the MMC5 split screen. Had to debug the game to figure out how it saved game data so I could cheat my way to the ending. https://forums.nesdev.org/viewtopic.php?t=12764

[glacial turtle]

I think he was responding to my faux-naive musing that the people who made the videos about how many of these tests MiSTer passed, and failed, would make a new video on how MiSTer now passes them all and then some.

[steep yew]

He knows. He is trying to troll me. But I’m on to him 📸

[dreamy marten]

https://tenor.com/view/meme-point-pointing-you-penguin-gif-13403813917065574318

[dreamy marten]

Haha thanks. Yeah seems like a slim chance unless he can tweak the test for his emulator to win again.

[glacial turtle]

He is in communication with Kitrinx, so I expect any new tests he will run by her rather than actively trying to trip her up without warning.

Is good he removed that dud test

[fierce estuary]

he added a new one because we told him about a new thing I found

so I (and I assume now he) are the only two emulators that pass it

im currently stuck on this stupid rom

there's supposed to be flicking garbage behind the 1 next to Marth's quest

but there isn't

and it's upsetting to me

you can see it drawing here

(I think) real hardware is supposed to do it

the core before my changes had some garbage there

but now there's no garbage

😡

[steep yew]

The people demand garbage

[fierce estuary]

the rom does nothing, it's just some poorly made demo from 15 years ago

but I love those, because they accidentally test all kinds of things

I have a whole collection of horribly broken roms

I pass ever single test but not this garbage rom

[dreamy marten]

I was implying that you had the ability to tell a whole story with a 1 second gif. You have a reputation to keep up.

[steep yew]

I try not to shit up Kit's threads too much. I save that stuff for waka's threads

[dreamy marten]

This thread belongs to someone named batty.

[glacial turtle]

Less said the better there, and yeah, let's keep this channel on topic, especially when the boss is cooking.

[supple ermine]

Well, I mentioned the Atari and Intellivision collections that used a NOAC. The NES patents expired and the games are licensed.

[timber lava]

Yes they were be close soon 🙂

[supple ermine]

It does bring up the issue of emulating the ways in which NOAC doesn't get things right

[main narwhal]

nah. the NOAC is just an obsolete hardware emulator at this point

[fierce estuary]

https://imgur.com/a/gPWDyIM

that's a NOAC

real hardware is kinda mid for accuracy https://github.com/100thCoin/AccuracyCoin/issues/20

[jade zenith]

i've heard about the krzysiocart before, is it any good?

[fierce estuary]

no idea

[urban meadow]

Time to make the core fail the same tests as real hardware to make it more accurate.

[main narwhal]

This is from testing a Retron 1 HD... it's a NoaC that takes carts (even runs an Everdrive N8)

funnily enough it ranks higher than an NT Mini Noir 🤣 but a bit less than the RetroUSB AVS

so yeah - why bother implementing that... if you want the palette it has you can create a custom one.
I suppose you could arguably have an option to break the audio into having reverse duty cycles...but why

[dry lynx]

First we had the dot core - next, the garbage core

[covert crypt]

But aren't analogue products 100% accurate, 0% emulation?!!

[dry lynx]

Analogue products aim to be 100% accurate to original hardware. MiSTer aims to be 100% accurate to original hardware even when original hardware isn’t 100% accurate to original hardware

It’s wild to me that people are still figuring things out about the NES after all these years

[half pasture]

Is the Mega Sg able to run Paprium through krikzz flashcart yet?

[dry lynx]

This is a Nintendo channel. Step away from the keyboard

[glacial pecan]

Can we have a accuracy switch?
Like 90%, 100%, random% or something like that

[half pasture]

I would give them names. Like “Retron”, “Analogue”, “NES”, etc

[covert crypt]

You have to put money into your MiSTer for the Analogue setting though

And it doesn't get turned on until Q3 of the next year

[cursive sable]

No problem, just run that ROM in Nesticle 🙂

[blissful prairie]

That's an acceptable tradeoff for finally nailing Metroids and Micromachines 😉

In the meantime we will suffer through Martin's glitchless quest.

[fierce estuary]

no, it's not good

mysterious regressions are never okay to ignore

[glacial turtle]

What a weird issue. Once it is figured out maybe worth the 100 coin man making a test for this garbage square issue

[vivid roost]

could it be one of those things thats different on different NES's?

[glacial pecan]

Rfe: Core setting for specific console serial compatibility

[vivid roost]

it does seem like kind of a glaring bug for the dev to of ignored, even in a demo

[dry lynx]

In we trust

[fierce estuary]

I actually found the issue with it, it was both dumb and really bad at the same time

it's ironic that all the tests pass with an issue like that

but reads/writes to the ppu were happening a cycle early and that stupid demo was literally the only thing that caught it

I had broken it when I was experimenting with something as I did this refactor

[covert crypt]

I think this says a lot about "accuracy"

[fierce estuary]

I mean, it was de-facto accurate

because read timing only impacts a few things which are mostly things being done wrong, and the code I wrote was compensating for the difference in read timing so the outcomes were the same

that test rom was different though because it's mapper timing was external, so it couldn't adapt to the local read/write timing

remember the PPU is supposed to be able to handle 4 different alignments so read/write timing is a bit flexible

[ripe shell]

I was going to do this for shits and giggles - inject accuracycoin into this and see how many tests are failing
https://www.gog.com/en/game/konami_collectors_series_castlevania_contra

but it seems like roms are packed inside of .exe and hex editor isn't helping me to determine where's the CV or Contra rom in it

oh well

[fierce estuary]

not all things need to be accurate for every game

like the things mister failed when he made that video didn't really impact anything

[covert crypt]

Oh well, just going off you saying it was really bad, it's interesting that it still passed all tests

I trust your judgement of whether things are really good or really bad when it comes to the NES core!

[fierce estuary]

one of the things I implemented in the wild impacts literally one single pixel on one screen in one known game

and it's homebrew

(8-bit stein's gate)

can someone give me a little clip of this on real NES? or at least a picture of when the corruption under the 1 is showing

[mint lichen]

This what you want?

[fierce estuary]

yes thank you

[half pasture]

Core is more than perfect now?

[cursive sable]

Perfecter

[fierce estuary]

Well, it's more consistent than real hardware

So technically a better tool for speed runs

[glacial turtle]

That's a good point, removes potential weirdness from aging hardware and model variance

[pseudo phoenix]

I wonder do all these core perfections (potentially) affect game glitches? Like solve game bugs as a byproduct

[clever scarab]

Likely no, as the goal is to match the original hardware as much as possible so all changes are done to do that.

If it fixes bugs present on real hardware then that’s a problem

[half pasture]

Thanks, Robby AI™ !

[clever scarab]

you’re welcome

[gray plaza]

Well there are some bugs that are harmless. I mentioned this one before with PPUCTRL bug, because it's practically harmless and only occurs on some CPU/PPU alignments I don't think anyone would shed tears over this not being there.

SMB1 is the one game I think off the top of my head where it comes up, there's a good explanation in 100th Coin's SMB3 in 5μs video

[cursive sable]

The goal is still to replicate what the real hardware would do, even if it's a bug that causes stuff like that

It's trying to be as accurate as possible, not fix things

And bugs like that actually can effect things like speedruns, especially if they rely on glitching the game in a predictable way

[clever scarab]

I will shed tears, I will miss those little wayward pixels 😢

[gray plaza]

... I shall retract my statement and hand you a handkerchief

[hasty trout]

The first rom hack I ever made was to fix that corrupted line bug in SMB1 actually!

It kept popping up on the hidef nes and it was annoying me

[clever scarab]

I honestly had no idea about that bug lol

[gray plaza]

haha nice

[gray plaza]

Yea I don't know other games that may trigger it, but there is a test ROM for it

[hasty trout]

Zelda 2 does, that much I remember

[hasty trout]

No idea about others though

[gray plaza]

Does it use something similar to SMB1 for its NMI?

[hasty trout]

Yup it’s very similar

[clever scarab]

I think it’s really cool how much y’all know about the NES. It’s an impressive dedication and I respect that.

[gray plaza]

It took years and I don't even think I know that much haha

Like I know some special hardware and some oddities about them, like the NES Test Station having 4K RAM compared to the usual 2K

[glacial turtle]

What mappers did you all add support for Lain? I will make a note on the sheet

[gray plaza]

40, 81, 142, 144

[glacial turtle]

Nice one, thanks. There any others you are looking at right now?

[gray plaza]

Nah, the problem is a lot of the mappers are either multicart or FDS conversion mappers. A few could probably be added without too much hassle but it might not be worthwhile when you can just play the FDS image.

Like for instance Mapper 103 has a horrible banking config.

[cursive sable]

Or horribly complex like the flashrom ones for homebrew games

[pseudo phoenix]

Watched the video. This guy is amazing. All these minute details are fantastic.

[supple ermine]

Well the idea is to do the Atari/Intellivision. NOAC comes up because that's what they used. I assume if you emulate the mapper with a regular NES the games will still work.

[gray plaza]

There’s an explanation here as well: https://www.nesdev.org/wiki/PPU_glitches#Affected_registers

[fierce estuary]

the glitches like that only happen at certain alignments

they are never good, so emulators rarely provide that alignment except as an un-recommended option

it's because nintendo was shitty and M2 goes high before the bus is actually stable so continuous writes can write open bus to a register for long enough to trigger things like oam corruption and etc

well m2 goes low technically

it's active low

[shy plinth]

Does the American NES Zapper via Snac work on the Famicom/JP versions of Duckhunt and Hogan's Alley?

[urban meadow]

I don't see why it wouldn't.

[cursive sable]

Isn't the port entirely different?

[fierce estuary]

I don't see why it would

the register reads are different for famicom port vs controller port zappers

iirc

[cursive sable]

Is there any sort of SNAC support for the Famicom's expansion port that the gun would have plugged into? Or does the NES core only support the NES controller ports for SNAC?

[fierce estuary]

I think it's mostly wired for nes atm with the exception of glasses and some other stuff

I wonder if this works properly on real hardware

[sharp geode]

i'm a little bit confused, what's the current version of core, despite on accuracy tests and etc.?

[hasty trout]

Looks like it works

[fierce estuary]

I have a mystery how that can work and also this test can work at coordinates 30, C4

[covert crypt]

When the moon hits your eye like a big pizza pie, that's a moire

[mint lichen]

idk if you want vid of those coordinates

[fierce estuary]

I think I see

[cursive sable]

Glasses?

[fierce estuary]

active shutter 3d glasses

[cursive sable]

I know the Master System had those, I don't recall seeing those for Famicom

Neat

How do you even use those on Mister?

[hasty trout]

I’ve been meaning to get a pair of those at some point

[cursive sable]

Heh, I am guessing my Nvidia 3D Vision 2 Glasses or the active glasses for my Samsung tv won't cut it XD

[blissful prairie]

Rad Racer in ultra mega 3D

[fierce estuary]

the japanese ones were active, the us ones were red/blue

[blissful prairie]

In Europe we also had those red/blue glasses for Rad Racer. I cannot remember that we used them a lot. I really loved Rad Racer. Was really disappointed by Rad Racer 2 when I got to play that a few years later.

[fierce estuary]

I think im finally done with this PR

that was a lot

[hasty trout]

Amazing work 🥳

[fierce estuary]

everything somehow works exactly the way it's supposed to

[granite pivot]

3d glasses adapter plugs into the expansion port and fits into the famicom snac adapter. Adapter has a phono plug for the glasses.

[cursive sable]

There is a famicom expansion port snac adapter?

[granite pivot]

US also got the red and blue, but I have the FC version which uses the shutter glasses. JP FDS 3D Worldrunner (forgot the JP name) actually uses red and blue glasses, as well.

Misteraddons sold it. Not sure if they still do, but it has an hdmi style plug and also fits in the reflex adapter

[half pasture]

Waiting for the hyper-accurate release of the NES core so I can play SMB1 in hyper-accuracy mode!

[glacial turtle]

Hype! Unstable nightlies incoming?

[hasty trout]

Hopefully soon!

[fierce estuary]

yeah I need to fix a small thing, it will need a lot of community testing, cause I changed.. a few things https://github.com/MiSTer-devel/NES_MiSTer/compare/master...Kitrinx:NES_MiSTer:more_fixes?expand=1

[granite pivot]

The passthrough on the 3D adapter actually works for stuff like the hudson joycard with SNAC, but the controller gets detected as a P1 controller.

[steep yew]

Good god

[fierce estuary]

I told you those last couple of tests needed a refactor

[glacial turtle]

Do you know if 100 coin man is working on more tests or he drawn a line at this point?

[fierce estuary]

Probably

We were talking about odd even nmi timings

[glacial turtle]

Can a test be made for the weird garbage square issue you found? I imagine that could trip up a lot of emulators

[graceful field]

Oh did this bring in some of the earlier work around ppu alignment?

[fierce estuary]

Alignment was the one thing I didn't bring over

I'll do that separately

[steep yew]

Is that the boot variance stuff?

[fierce estuary]

Yes

There was too much in this pr already

[fierce estuary]

here, please test this a lot

save states, weird mappers, pal stuff

video settings

this is release candidate, if it passes the initial round in here ill post it in test builds

a lot changed

[glacial turtle]

Oh, are there new mappers?

[main narwhal]

new PPU

[timber lava]

Thank you ❗

[tepid umbra]

NES is my favorite core, thank you for your dedication Kit.

[timber lava]

🙂

[blissful prairie]

I will do on the weekend. I guess that this is just for the accuracy fixes but not yet supposed to show the NTSC borders?

[half pasture]

SMB1 runs flawlessly and hyper-accurately. Release wen

[covert crypt]

I'm gonna sleep so well knowing that SMB1 is now accurate.

[mint lichen]

If you have anything more specific you want tested, let me know, otherwise I can run thru a lot of games I know with save states since you mentioned that

[clever scarab]

Can you please test Battletoads, need to 1CC it though

[blissful prairie]

Japanese Battletoads (Famicom) is recommended for a 1CC. Bugfixes and much more bearable difficulty.

[livid leaf]

https://tcrf.net/Battletoads_(NES)/Regional_Differences

A surprising amount of differences.

[dire lodge]

Ooohh, exciting~

[grand jewel]

I finished Batman - The Video Game (JPN version), Rockman 1 + 2 + 3 + 5 earlier with the new RC core.
Verified some stages on Insector X, Akumajou Densetsu & Micro Machines too.

So far, so good.

[livid leaf]

No Rockman 4?

[grand jewel]

I was missing some time, and I don't speedrun it (sole one I never had during my childhood).

[hasty trout]

Nice 😁

[livid leaf]

Okay, just seemed weird to skip from 3 to 5 like that.

[grand jewel]

The save states function are okay on the games I cited (from a previous version, creating a new state state file, rewriting an old save state file).

[grand jewel]

I know much better 2,3,5.
And I wanted to see if my 'muscular memory' was okay with the core.

It's completely intact.

[fierce estuary]

it shows them, it's under the "overscan" option in audio/vido

[fierce estuary]

smb is really simple, it shouldn't have changed

most games shouldn't have changed

testing things I wouldn't think to test is best, japanese games, games with complicated mappers, pirate games, etc

I was really careful to try to not break old savestates

[glacial turtle]

When you mentioned "weird mappers" are there specific ones that need looked at? Are any new ones added that should be tested?

[fierce estuary]

the complicated ones that count address line changes

mmc3, mmc5, that really big pirate mapper I can't remember the name of

[grand jewel]

I tried some PAL games too, like Asterix, New Ghostbusters 2 and Hammerin' Harry.

[fierce estuary]

I technically fixed a bug in MMC1 too that caused a japanese game to crash

Elite is the most challenging of the PAL games and that one was working okay, so they are probably fine

[lean stag]

In Rad Racer 1 & 2, i see a flickering line in the right part of the screen.

[grand jewel]

Are you sure it's not inside the overscan part ?

[fierce estuary]

check extra sprites

does it look like this? https://www.youtube.com/shorts/-4_ZK621VEQ

[lean stag]

Yes, it also appears with Extra Sprites enabled.

[fierce estuary]

I see it in this real hardware video too https://www.youtube.com/watch?v=3me85pqMwAQ&list=RD3me85pqMwAQ&start_radio=1

[latent breach]

MiSTer wins again

[fierce estuary]

those are 2006 writes delayed interfering with the natural increment

that's probably correct behavior

someone with an everdrive can verify or not

[lean stag]

Yes ok, but in the rbf from 25092025 is no flickering line.

[fierce estuary]

that's fair, there was no 2006 write collision behavior implemented then

blizzard did some wild stuff with the writes in this game

if someone uses a crt, make sure it still works okay with the video timings, and with the overscan set to "borders" or "everything"

I changed the video module so it shouldn't lose sync when loading new games

because the ppu generates it's own signals now

[hasty trout]

Working perfectly on my CRT with overscan set to Borders and Everything

looking great so far 😁

[fierce estuary]

great, the borders might actually be nice on a crt

that's what it was meant for

to hide the edge masking

if someone with a real nes could try rad racer just to verify that would be great

because mesen doesnt do it, but mesen is not quite as accurate

[hasty trout]

Looks 100% spot on to me

same glitch in core as my AV Famicom

[fierce estuary]

great, that's a win for mister

[mint lichen]

@fierce estuary idk if this is a bug or not with the Overscan setting, but loading a save state from the old core changes the Overscan? It is showing everything on the old core but then is cropped after loading on the new, even as I show it's set to Everything before loading it. Hopefully I'm explaining it right, I tried to show everything https://streamable.com/4wn0h8

[ripe shell]

sweet, I don't keep losing sync on OSSC

[fierce estuary]

I had to expand the setting by 1 bit, so it's in a different position, it wont match the old core's overscan setting unfortunately, so you'll have to reset it

[fierce estuary]

great, that was annoying me too

[mint lichen]

Okay, cool. I did notice a difference with that game in that clip tho. This is MMC5 Rockman 4 Minus Infinity. On the second screen, showing the devs site at 32 secs, you can see on the old core it's fully black, and on the new there is a white bar on the left and I think there was a single pixel line on the bottom at 1:01

[fierce estuary]

the white bar on the left is a thing a real nes does, it's in the overscan so you dont see it on real tvs

[hasty trout]

Yeah, I have a Rockman 4 Minus Infinity repro and it does the same thing

[mint lichen]

But it isn't present on the old core

[fierce estuary]

the border doesnt draw on the old core

[hasty trout]

It’s accurate behavior

[mint lichen]

Got it, just reporting a difference I noticed

[fierce estuary]

see it's drawing a white border

[fierce estuary]

thank you

that one goes in the "more accurate" column

fun fact, at the retro con I was at a few weeks ago, I met Infidelity, the guy who made that hack

wait

is that one infidelities?

maybe it isnt

[mint lichen]

No, puresabe I believe

[fierce estuary]

the mario all stars is infinidelity, and the legend of link

this is kinda awesome to see these small but noticeable differences

[glacial turtle]

Is brilliant work, and no regressions found (certainly so far)

Has anyone checked the music player yet?

[fierce estuary]

hmm, it plays but looks weird

some kind of alignment thing, once in a while it looks good

@mellow dawn any ideas? I can fix it real fast if you give me a hint

this is the mmc1 fix btw, doing this in the old core would freeze it

the look option on a blank space

the single flickering pixel on the last screen of this also took a lot of effort to figure out

speck of dust indeed

[glacial turtle]

"Periphery" now "Peripheral"?

[fierce estuary]

that was from like, years ago

[glacial turtle]

Yeah and weirdly slipped through the cracks, I had missed it until I read that ticket and checked

[fierce estuary]

I have to check what's going on, my builds went from 26 minutes to 30 minutes when I merged Lain's latest code

[cursive sable]

Wait, is that NSF player built in?

[fierce estuary]

yes

[cursive sable]

I never noticed it had NSF listed under filetypes you could load

Wish I had an everdrive or something to test it on a real NES, wanted to hear this One Winged Angel NSF but I think it sounds different from the youtube video, but it's also from 2013 and the creator appears to have been using some PC app to play it so it might not have been tested on real hardware anyway. Going to assume that the app was inaccurate... or maybe my ears just suck and both are exactly the same.

[timber lava]

Stein;Gates is a new (seeing M2 logo) game ? never knew about it.
Thank for checking this issue and fixed it!

[cursive sable]

IIRC it's a super popular visual novel from a few years ago

The original I mean

So yeah, would have definitely been well after the NES's retail time

[tepid umbra]

It's also a super popular anime series.

[fierce estuary]

yes but it was developed on an emulator and has many problems

[timber lava]

I see. It could hangs on real hardware or it is just graphical issues ?

[fierce estuary]

it doesn't hang but it has glitches and shaky text

[tepid umbra]

Steins;gate elite on switch had it packed in as a bonus game and it wasn't very long nor very good tbh.

[cursive sable]

Looking it up, apparently it was included as a bonus/side-game among re-releases of the original on modern platforms?

I guess I can see why they don't care about it working on real hardware if they never produced actual carts and only intended to include it with the main game as a bonus to run in emulation

[glacial pecan]

While I like what they are doing, how much can you trust a company creating emulators/ports if their own emulation seems inaccurate enough to not produce any problems?

[cursive sable]

I mean, if the purpose is just to run that one game, I doubnt they are going to care to make a super-accurate emulator. It's only purpose is to run their one game

[ripe shell]

M2's main forte seems to be Mega Drive first and foremost

[cursive sable]

I thought M2's main forte was NVMEs?

XD

[ripe shell]

-# heh

[cursive sable]

Hey, it was either that or a Bass joke

[urban meadow]

I thought M2's forte was the 3DO. 💪

[glacial turtle]

Let's keep the banter elsewhere while there is a request for testing going on for the biggest update to the core in years

[fierce estuary]

can someone help me verify this bug? https://github.com/MiSTer-devel/NES_MiSTer/issues

[hasty trout]

Which bug? The link seems to just go to the list of issues

[fierce estuary]

oops one sec

these bugs seem to be addressed by this:
#416
#412
#375
#367
#366
#364 - Maybe
#359
#235
#192

https://github.com/MiSTer-devel/NES_MiSTer/issues/370 this is the one I needed verified

[timber lava]

#364 I couldn't reproduce

[fierce estuary]

thanks wark

[timber lava]

I have the rom with md5sum a6618b437fb720b275f138081058075f (NES2.0 header) that work mention on the issue. So maybe an old demo or not good header for the rom having issue ...

[fierce estuary]

maybe so

either way it seems like not a bug anymore

if it ever was one

I tried to evaluate sword master and dont notice the missing channel

[mint lichen]

They posted in here the other day about something else, could tag and ask if it's still happening

[fierce estuary]

@sharp geode

[timber lava]

seems ok for me but I'm not good with sound

[fierce estuary]

it might be part of the save state

I think it doesn't save all the audio info perfectly

I have a really hard time telling

[graceful field]

There are definitely scenarios in the current released core (not your test core) where loading a game from the savestate will initially have missing or odd music, like missing parts of the song, so definitely could explain it. At the same time, at least in my experience the audio fixes itself after a few seconds.

[fierce estuary]

I think that's why it is that way, to simplify the save state it ignores some audio stuff since it tends to be self correcting

@rugged storm actually implemented that so im not 100% on the details of it

I would consider it a don't-fix unless there's an actual problem with the APU missing something or timing issues

[hasty trout]

It does sound like a channel is missing still, but it’s a pretty small difference. I’ll have to learn this game so I can test it on my Famicom too

[fierce estuary]

that's the last phase of the final boss so it's not that easy

but a savestate earlier would make it easier to test

possibly pausing then unpausing might help

[mint lichen]

There's a code to start on the last stage, with other codes you can probably get there super fast

https://gamefaqs.gamespot.com/nes/587675-sword-master/cheats

[timber lava]

This demo is working good with PAL settings but on NTSC it is only blackscreen with sounds
https://www.pouet.net/prod.php?which=91816
On the forum it is PAL demo but wonder if someone could test on real NES NTSC to see how it behaves
https://forums.nesdev.org/viewtopic.php?t=24030

[near cloak]

Robby sent me

[mint lichen]

On AV Fami, black screen with music

[timber lava]

thanks 🙂

[fierce estuary]

PAL stuff is probably somewhat inaccurate

this is known

the next thing im going to do is try to add variable PPU alignments which is required for accurate PAL stuff

it was too much for this single PR though

does that video I posted have the missing channel?

[hasty trout]

can't seem to hear any audio from it, I tried on my phone and computer

[swift temple]

I hear nothing

[fierce estuary]

ffff

sorry

I forgot I muted that

ugh

damnit

lol that's in PAL 😭

[swift temple]

Is the blue dude supposed to have sound when firing the fire bombs?

The first vid also didn't have noises there

[fierce estuary]

here

jeeze

really failing at this capture

removed evidence of failure

[twin siren]

i was told this was a place to be

[fierce estuary]

you were thinking of #1103404843512631357

[twin siren]

thanks

[hasty trout]

yeah still missing there, compare at 15 seconds in video above to https://youtu.be/WVEposcLZcE?t=70 this at 1:10

[fierce estuary]

and the real game on a real nes has that channel?

[hasty trout]

not sure, I'll try to find out

[fierce estuary]

is the nsf from japanese or english?

nsf's arent really nessesarily accurate

[rugged storm]

it's been 4 years, I barely remember what we left out to make it fit well. We can always include more registers if required.
Not much work and typically compatible if we use one of the free spots, but we would need to know what to include. Just adding everything is too much.

[fierce estuary]

I don't think it's super important that audio restore perfectly as long as the line counter does, which it seems to

[sharp geode]

in my video, which is recorded december 2023, missing one sound channel, it should be some basses

[fierce estuary]

that's the USA version?

[sharp geode]

it should be like this https://www.youtube.com/watch?v=s3y8la4BxxU

in USA, and maybe in JP version

[fierce estuary]

I just have to make sure the real system does this

[sharp geode]

yes it does

[mint lichen]

I can get there on hardware to see

[sharp geode]

and i don't remember, which version of core was

[hasty trout]

that video is past the point where the issue appears

it's right as the music starts

[mint lichen]

Okay, I have that too, it was just a bit too big, one sec

https://streamable.com/e5zi1g

[ember bone]

What games need to be tested the most?

[hasty trout]

Nice thanks, does sound like real hardware is different

it has the channel

wonder why it's missing from the MiSTer core, that's interesting

[fierce estuary]

that is interesting

ill mark that one down

im going to have to step through it frame by frame in mesen to see exactly where the $4000 writes are failing

[mint lichen]

Kitrinx said "testing things I wouldn't think to test is best, japanese games, games with complicated mappers, pirate games, etc"

[fierce estuary]

the core in test builds is the same I posted here earlier

opening it to more people to test

[tame flame]

Nice!

[fierce estuary]

you can turn on NES behavior in Advanced options to pass column 15

the one in 19 he removed becacuse it wasn't a good test

[tame flame]

Oh, there's a new test rom?

[fierce estuary]

yes

[tame flame]

Amazing

[fierce estuary]

good luck passing all those on your real nes, pff

[heavy dust]

I should try this on my famicom and twin famicom both

Are Everdrives likely to cause some of these to fail vs actually burning a real rom on a real cartridge like we tried to back in the 90's?

Always been curious.

[fierce estuary]

yes

[hasty trout]

#1091056042667937944 message

There's a built RBF for the Everdrive N8 Pro that fixes 2 of the tests

but yeah it does impact them

[fierce estuary]

in all fairness, a real nes does pretty well, it ranks 2nd or 3rd in accuracy

[orchid oracle]

More NES than NES.

[fierce estuary]

well, the everdrive is to blame mostly

but also there's a variety of PPU and CPU variations. the best one is the early G models

this test is made around that

earlier famicoms had some undesireable bugs in the PPU especially

[heavy dust]

There's enough revision differences in real consoles that I figure we can just view most existing MiSTer cores as not necessarily inaccurate so much as just another minor revision.

But I'm also wrong about most things.

[fierce estuary]

there's things that can be just alignment/revision difference and things that can be just broken

like for example, if the rendering being enabled is not delayed by at least three cycles, battletoads will freeze on level 2

the target for mister, if you care, is early model G

it's the most compatible and the one you probably played on as a kid in the US

if you were a kid in japan maybe you had an earlier one, and if you were a kid in europe, your life experience is invalid

[timber lava]

Mapper Topic - I remenber this mapper (NES 2.0 Mapper 552) is not implemented https://www.nesdev.org/wiki/Taito_X1-017
It is uses for those retail games
SD Keiji: Blader
Kyuukyoku Harikiri Stadium 1989 Edition
Kyuukyoku Harikiri Stadium III
Kyuukyoku Harikiri Koushien

[pseudo phoenix]

NES rules. @clever scarab back me up here

[clever scarab]

This man speaks the truth!!!!

[fierce estuary]

what about it?

[heavy dust]

If you were a kid in Japan with an earlier one you hopefully upgraded to one with non shitty buttons.

[timber lava]

During my tests of new test core, on game SD Keji Blader it shows gray screen and just remember this mapper is not implemented.
This mapper was RE not many years ago (2020).

[fierce estuary]

oh I see

it's a retail game with an unimplemented mapper

[timber lava]

It is an RPG developed by NMK and published by Taito.

[fierce estuary]

can someone test this on real hardware?

[fierce estuary]

supposedly it will flicker around a lot when you press the A button

[dim relic]

AV Famicom with NES RGB mod and everdrive N8 Pro.

[fierce estuary]

thank you

[half pasture]

Dat CRT 😍

[mint lichen]

@fierce estuary think I spotted an accuracy improvement? In Tom & Jerry on several of the title cards there is a little glitchy thing on the left side, which is present on hardware, but you can see it's not there in the old core. Near the bottom left of the big triangle, I show old core then new. https://streamable.com/hrgruo

[fierce estuary]

cool 🙂

[fierce estuary]

you tested it on real hardware?

it's a question of it extends outside the border or not on real hardware

it's hard to tell

if anyone has a good recording setup and a nesrgb for clean pixels that would be interesting

[glacial turtle]

Maybe one of the youtube men could help there

[fierce estuary]

fiskbit is questioning if the timing is right or not

[hasty trout]

I’ll record it now, was on a work call

Hmm I don't see it here

I thought maybe it was being cropped out by the RetroTink but I messed with the cropping and didn't see any flickering

[elfin flume]

If you have left edge masking happening under Audio & Video that is enough to hide it. Assuming that was a capture from the core.

[hasty trout]

Good to know! I made sure it’s not being cropped out in that AV Famicom capture. I do have a rev H CPU/PPU if that affects this.

[supple jasper]

I'm happy to see the dumb borders in overscan settings now. I know I said forever ago that the mask left/right side settings did nothing, it still doesn't, but it's all rolled into the overscan setting. Kitrinx told me at the time that it's not important enough to rework and I absolutely agree with her but it's one of those mental ticks in my brain telling me this is closer to the actual hardware now. Thanks for putting in work to pass tests that don't really matter all that much

[fierce estuary]

that's fine, I think it appears in the border

so it might be extending one pixel too far

[mint lichen]

I just have composite AV Fami but you can see it

[hasty trout]

Huh, yeah! I wonder why it isn’t showing on mine

[mint lichen]

Are you trying the JP version? I don't remember if US has it

[fierce estuary]

nesrbg apparently breaks it

[mellow dawn]

Looks like something went wrong with copying the chr data from prg. The letters are there but not the numbers (maybe some of the symbols). https://github.com/MiSTer-devel/NES_MiSTer/blob/master/rtl/nsf.asm#L96

[topaz moon]

ldx #3                  ;load CHR: 128 tiles = $800 bytes

[mellow dawn]

Is this using NES or famicom writing to ppu quirk?

[fierce estuary]

umm, good question

famicom

[fierce estuary]

it's random if it shows up right or not

[mellow dawn]

I suspect it works with NES

[fierce estuary]

it likely does

is it just loading this rom?

[mellow dawn]

It is not waiting to copy the characters to chr ram

[fierce estuary]

ah I fixed the writes to be on the second cycle now with the address latching and everything

[mellow dawn]

It is based on Loopy's powerpak nsf player which would mean it is always after the block to writing to ppu has past.

[fierce estuary]

famicom behavior allows this

[mellow dawn]

Or whichever one allows immediate writes

[fierce estuary]

which is the default

[mellow dawn]

Oh, and you were already using famicom?

[fierce estuary]

yes

does it use a mapper?

I changed mmc1

[mellow dawn]

Hmm. Maybe something else is corrupting the ppu address

[fierce estuary]

many things can corrupt the vram address now

if done wrong

[mellow dawn]

Very likely to be something like that. I don't know all the things that need to happen to avoid messing it up. The asm probably needs to be fixed. I would not hold up the pr for it.

[fierce estuary]

it still plays the music so ill leave it for now

if you ever feel like it maybe dig in

[mellow dawn]

yep

[half pasture]

https://tenor.com/view/dew-it-galactic-republic-palpatine-gif-21847982

[hasty trout]

Sorry missed your response somehow, I tried the JP version too and it’s not there. I think Kitrinx is right about it not showing up on RGB modded systems

[fierce estuary]

fiskbit told me the NESRGB doesn't use the vram address to draw it but rather just guesses and uses the color in the first palette address

so it's not accurate

I didn't know that detail personally

[hasty trout]

I didn’t either! Interesting

[mint lichen]

I'd seen it thousands of times doing speedruns of the game on console, so I remembered it and saw it on the new core and randomly checked the old. Cool it shows up now.

[ember bone]

Playing Uncharted Waters the past couple hours. At the point where I got my five galleons in Seville (the speedrunning way), no issues yet.

Usually this game loves to give trouble to flash carts

[cursive sable]

IIRC those RGB mods require you to de-solder the PPU and solder a socket in it's place that will go between the board and PPU, I think it just... mostly does it's own PPU implementation when outputting over RGB? The HDMI mods even make you do this to the CPU. So I would not be surprised if some quirks are different with these mods if they are indeed replacing parts of the PPU/CPU like that

Oh, I see the modern ones have an entire daughterboard you place over the ppu

[dry lynx]

It would be pretty amusing if people spent years trying to match hardware accuracy only to realise that the specific hardware they’re emulating is actually just everdrive + HDMI mod, and that the weird behaviours don’t actually exist on the original unmodded console with an official cart 😆

[timber lava]

Like some Pcengine RGB mod.

[tepid umbra]

Chasing the dragon.

[dry lynx]

We’ll go full circle when people have composite IO boards and when cartridge inputs become a thing

[gleaming crag]

Just curious if this behavior happens with real hardware too. I’ve been playing Tiny Toon Adventure and on stage 1-2, there’s a side strip flicking on the right side of the image

[ripe shell]

yeah that's regular overscan garbage

[tepid umbra]

Guess I'll play fe1 and 2 on the new core since I'm the only one who likes those games on nes.

[blissful prairie]

Fe?

[tepid umbra]

Fire emblem.

[dry lynx]

Absolutely great game!

[acoustic tangle]

Just read the news

Suck on that, TriCNES

Also, suck on that, accuracyCoin dev

[mint lichen]

Think I spotted another tiny improvement in Addams Family. Little dots on the top right of the HUD, I assume due to a screen split. Three clips, there on hardware, old core, new core.

[mint lichen]

Is it known that a save state made on the new core can break on the old core? Had it happen last night when I wanted to check something in Little Samson on the old and loading the new core save state broke the game. Idk if it's intended that the save states only work going old to new

[fierce estuary]

i'd expect that, I was pretty careful to map the old save states to new values, but the new ones don't really map to the old ones, they have more info

[mint lichen]

Cool, kinda what I figured after you said you added a bit to the overscan setting that there was more info that wouldn't match. Obviously won't be an issue when the old core gets replaced.

[fierce estuary]

it should (hopefully) migrate cleanly forward

[foggy bramble]

two tests in the accuracycoin suite are failing when snac is enabled. is that expected? page 13 (dma 4016 read) and page 14 (controller clocking). thanks for all your hard work @fierce estuary, it is amazing!

[fierce estuary]

that doesn't really surprise me, 4016 is a controller read, snac kinda does things the test doesnt like (I think it wants no p2 controller and stuff like that)

im guessing because how snac looks electrically it looks like there is an unreponsive controller connected

it's good to note, but it shouldn't impact any behavior of the system

[foggy bramble]

i am using snax64 with only one controller plugged in. so that might be the reason. it works fine (128/128) when using usb (reflex adapt)

[fierce estuary]

I'll look, but I suspect I can't really fix that because of how snac and the test work

it's just a false positive

[gray plaza]

I know some tests fail if you have a SNES controller plugged in for example

[regal perch]

kid dracula might be good to test. only nes game i tried that behaved different across 3 dif mister fpga set ups

1 worked perfect, 1 worked with graphic glitches and the third just spazzed out entirely after the first 10 seconds or so

not with recent. just in the past

[cursive sable]

You joke but that's not a bad idea. Would not have to worry about mappers or other such extra hardware with a cartridge port so you would be able to verify that the core system components are functioning as they should, as well as test how a game off a real cart with a real mapper functions vs a ROM of that game with the mapper on the FPGA, would likely be helpful for development. Though you would also need the carts in question that you are testing since an everdrive can add problems and would just be emulating mappers on itself then.

[tepid umbra]

I don't even understand what a mapper is but I still have fun playing the games.

[ember bone]

Sometimes that happens, yes.

[cursive sable]

I am probably NOT qualified to explain this but I'll try. The NES/Famicom maps the cartridge directly to the system's address bus. There is only 40KB of address space left to map so that's all a cartridge gets, it's split into 8KB for the graphics (called the CHR ROM) and 32KB for everything else, the code, music, etc (called the PRG ROM). This was very limiting, but at the time was fine for things like Pac-Man and other such ports nd prices of ROM chips were high, but it quickly became far too small for more complicated console games. Supar Mario World is likely it's limit. There was ways to try to compress an optimize data storage, but that can only do so much. The FDS was originally designed to bypass this limit in a cheap way but ROM prices started to fall drastically which made it moot.

If you looked at a basic NES cart you would just see three chips, two big ones that are the PRG and CHR roms, and a small one that is the NES10 security chip (the Famicom versions would be even lacking that security chip)

But even if you could now affordably put a 258KB chip on there, the NES would only be able to access 8KB of it as a CHR ROM, or 32KB as a PRG ROM. That's what mappers are for, they "map" different parts of the actual ROMs to the address spaces the 32/8KB of the PRG/CHR ROMs. So it could for example map 4KB of one part of the 256KB CHR ROM here, 2KB there, and 2KB there, and keep swapping them out on the fly depending on what the game needs, same for the PRG ROM. It can never map more than 32KB at one to PRG or 8KB to CHR but it it can map different chunks of the actual larger ROMS to those 32KB/8KB address spaces.

Some mappers also had additional enhancements on top of this to allow for more effects or functionality, though I know very little about that.

[tepid umbra]

So it's like the memory banks in a game boy rom?

[cursive sable]

I guess so? I am not familiar with those to be honest

Some could even map RAM to the ROM space, so you could render something in software for example, write it to a a SRAM chip on the cart, and map that to the CHR ROM.

The NES version of Elite did this. The NES's PPU if I recall has no functionality to "draw" things like a shape, or a line, or even a simple pixel/dot. It can basically only copy-and-paste what is in CHR ROM to the screen

And perform some modifications like flipping it

[tepid umbra]

The game boy can also only access a tiny portion of the rom but switches banks to be able to offer more room. When I worked on the space world version of dx violet I had to hijack additional banks for the bigger sprites.

[cursive sable]

Which would make a game that uses wireframe 3D like Elite impossible... so they actually rendered the game's visuals in software, wrote them as tiles to a SRAM chip,and mapped that SRAM to the CHR ROM

Not sure if this is a mapper thing but it also let you offset ome tiles, this let you simulate a parallax scrolling effect, Battletoads used this to great efect

Any NES game larger than 40KB needs a mapper for this, and there were many different ones that functioned differently and had different features

Though I definitely do NOT know about the details of those differences or how they worked at a lower level

Retro Game Mechanics Explained has a great video on the parallax scrolling

[mellow dawn]

2000 words: Super Mario Bros = Mapper 0 aka no extra circuitry/chips

Bandit Kings of Ancient China = Mapper 5 = MMC5

[tepid umbra]

I love that most of those chips are THT. Only the MMC5 chip is SMD. Modern devilry.

[mellow dawn]

The only smd chips at the time that I can think of are mmc3 and mmc5 and mmc6. Zelda (mmc1) predates it

Super Mario 3 and 2 however use mmc3

Even things like vrc6 used in Japanese Castlevania 3 that produce extra audio channels are not smd

[mint lichen]

@fierce estuary Noticed this in Celeste Mario's Zap & Dash, a hack that uses MMC5. On the new core you can see lines on the bottom right and left when scrolling, which I don't see on console or old. 3 clips comparing, console, old, new. Is it just overscan? I can clearly see the edge of it on my tv

[fierce estuary]

that's being drawn in the border area, so out of the visible frame

I assume they are just well timed palette writes

most of that would be hidden by tv overscan

but i'd expect maybe a smidge of it to show if it's happening on a real system

it's the area outside the 256 drawn dots

ill check it out

[cursive sable]

Mario 3 is the biggest example I can think of of a game with a lot of gargage on the edges of the screen intended to be hidden by overscan. Actually, did overscan even hide it all? Could have sworn I saw some of it even on CRTs back in the way due to the fact that it could scroll both vertically and horizontally when NES games were meant to only scroll one way so the screen could be duplicated to hide some of that.

[fierce estuary]

actually no

because the NES had borders in the overscan area, on most decent tv's the entire 256 visible dots would be showing, so you'd see the edge corruption

the left edge also corrupts but nintendo built in masking for it, but they didn't anticipate fancier games with 4 way scrolling like mario 3, so there's no right edge masking

there is in the core a setting to mask it though, if you find it distracting

[cursive sable]

Thanks, so I wasn't just mis-remembering that I saw it back in the day

[tepid umbra]

In final fantasy and dragon quest there is also a lot of corruption but on the top and bottom iirc.

[fierce estuary]

@mint lichen I think this is probably a minor mapper issue that is changing the palette address during overscan

the game doesn't do much there

you are sure that doesn't happen on an unmodified real console?

(nesrgb would break it)

[mint lichen]

The first clip was composite AV Fami, that's as much as I can see on my capture. I can try front loader too

[fierce estuary]

when you capture on av fami, does it show the border at all?

like in super mario bros on the first level it would show a little blue/purple (sky color) on both sides

[mint lichen]

Yeah, I had it stretched this far to see on that capture

[fierce estuary]

ok

thank you

[fierce estuary]

@mint lichen I think I found the issue

pretty minor

ill have a build for you in a bit

[mint lichen]

Sweet, so it was a regression?

[fierce estuary]

no

we just never drew the border before

wire [4:0] pram_addr = is_rendering && in_visible_frame ? pixel : (is_pal_address ? vram[4:0] : 5'b0000));

this line should have been this:

wire [4:0] pram_addr = is_rendering && in_visible_frame ? pixel : (~rendering_enabled ? (is_pal_address ? vram[4:0] : 5'b0000) : 5'b0000);

the vram address is only used for the border if rendering is disabled

but we never drew the border before so we didnt know it was doing that

those literals are also too small but oh well

[mint lichen]

So it just wasn't supposed to have "show that stuff" turned on? My very technical understanding, haha

I'll run thru all the stuff in that hack since I know it well and it's MMC5, I just noticed that right away starting it.

[tepid umbra]

I notice that the new core is way zoomed in compared to the old core, i take it the overscan settings changed?

[mint lichen]

Normal for me, I think Overscan stuff just covers up the edges you normally can't see, so not sure about being zoomed in

[fierce estuary]

basically the palette ram is only supposed to use the vram_address when it's not rendering, otherwise it's just 0

in the case of those lines, it was rendering but outside the visible frame, so it was using the vram_address instead of 0, because there was never a need to limit it before

[echo night]

Just noticed you have a build that allows us to see all the overscan area now!! Thank you!!! Can’t wait to test. Will try tonight! 😁

[fierce estuary]

it's not zoomed in but it defaults to 256x224 which is the intended viewing area, before I think it defaulted to 256x240 which shows the overscan

now you can choose how much overscan you want to see, it makes more sense in that order

[tepid umbra]

I have an external scaler do the cropping, so I ended up double cropped, but that makes sense now.

[fierce estuary]

the ppu generates it's own border and timing signals now

[half pasture]

When I thought the core was more than perfect already, now this!

Analogue bites the dust! 😌

[fierce estuary]

I don't think analogue has really been considered a contender for accuracy in years

[tepid umbra]

Yeah Kitrinx has really outdone herself. It's like early Christmas.

[fierce estuary]

im pretty happy with this

there's virtually no mysteries

no "oh this doesn't work quite the same" type roms, even the weird tests

[half pasture]

Until the next new test rom shows up

[fierce estuary]

sure they're finding new edge case behavior all the time

but I think there's not much left that would realistically ever impact a game that wasn't intentionally trying to trigger something

[half pasture]

Phantom System core wen

[fierce estuary]

I really want to implement tas again

[half pasture]

Thanks for everything you’ve been doing with this core, @fierce estuary

[fierce estuary]

how do I keep myself busy between core builds, you might ask?

[tepid umbra]

By shaking your head: #spicy-off-topic message 😅

[fierce estuary]

I dare you to tell me i'm off topic

@mint lichen there you go

that fixes two little bugs, one of them being your palette thing and the other a small thing with 2006/7 vram corruption

thanks, by the way, for being so thorough in your testing

[mint lichen]

Cool, I'll run thru some stuff in a bit. And np, I just happen to be good at noticing small stuff. NES is my favorite, so I appreciate all the work to get it so accurate.

[mint lichen]

Being able to play back a TAS would be really neat, but as a speedrunner, so long as it's made very obvious one is being played.

[fierce estuary]

I can make it obvious, but I don't see how it's any different than real hardware

you can run tas on real systems

also the way it works on mister, the rom resets when you load the tas script, so it can only be from the start

it would be hard to hide

I want to add button press indicator overlays too

so if people want to display their inputs they can

on automatically for tas

[glacial pecan]

that sounds like a nice feature every core should have

[graceful field]

I agree. Unfortunately there is a lot of FUD on anything not original hardware. There was even something about smb1 about preventing multiple key presses even though you could just crack open an original controller and do it too (glad nothing was done there). I think it is overblown, but as someone trying to push this more into speedrunning, I would just be worried about more concerns if being in tas wasn't clearly visible.

[fierce estuary]

I think the best way to deal with that is to put a controller display on the screen

that way they can just watch and see if you cheat that way or not

[hasty river]

Bonus you could make each controller look like the controller for the core, and you could use that same HUD for the controller mapping so like when it says "Press the button for C" it shows C on a genesis controller

[fierce estuary]

no, I would show it as just a series of dots

that's the least obtrusive

[hasty river]

waahh

https://tenor.com/view/lucy-weird-cry-dramatic-wah-gif-15506705

[mint lichen]

Sure you can cheat on hardware but you need external hardware, rather than just running a script.

And if you were to make a fake run, you could include resets to make it look more real, so starting from the beginning wouldn't matter.

At the end of the day, if someone really wants to cheat they will find a way. I just think there should be some kind of barrier to dissuade them.

[fierce estuary]

well mister you'd have to hide the start where you were loading the rom

or the script

people can just edit the core to do whatever though, just like they can modify real hardware

change the clock slightly or whatever they want

if I wanted to cheat that's what i'd do

replace the clock with a slightly faster one

[glacial pecan]

nah, it is unused new old stock which has not aged like yours

[mint lichen]

I'd want like a play button showing or something, like emulators do

[fierce estuary]

I can but there's no way to make it not simple to disable in code

[supple jasper]

You can turn all of those off on emulators. It's how they do video captures for youtube upload.

What Kitrinx said, if people want to cheat they will and they usually get caught. TAS/RTA communities have a pretty good honor system.

[mint lichen]

But the point is it's an extra step for someone to technically change it, rather than just hit play 🤷

[fierce estuary]

I think the way I implement it won't be ambiguous

but I can't stop people from modifying it

[supple jasper]

I don't believe BizHawk shows controller input or frame counts by default anymore.

[mint lichen]

Cool, yeah, I'd just want something to tell

[fierce estuary]

the loading screen will be the biggest tell

anyway I am reasonable confident that mister nes core will be the most accurate nes emulator that exists, for the time being, so I think it's fine for speedruns

if you find any more pixels out of place, let me know

[glacial pecan]

you could add out of place pixels when running a tas

something like dotcrawl

but more subtle

[main narwhal]

speaking of dots, what does the "debug dots" option rely on to decide when to show them?

[fierce estuary]

the dots represent when certain events occur in the ppu

mostly register writes and sprite0 hit

see the sprite0 hit drawn on the coin counter sprite at the top in SMB? 🙂

they are useful for getting a quick overfew of what is happening during a frame and when, so you can understand what might be impacting a glitch or bug or random smudge of color

see the rendering enable and the subsequent delay before it kicks in and the palette color turns into bg data?

they are a thing for me, and the bravely curious, I suppose

they correspond to a similar tool in mesen too, for comparison

mesen has really good debugging tools that are impossible on an fpga

ill give this testing another day or so and then merge the PR

[hasty trout]

Sounds good 😁 I haven’t found any issues yet

[fierce estuary]

tbf there weren't many issues before

[timber lava]

did you find the cause of missing channel on Sword Master ?

[pseudo phoenix]

I read most issues are visible glitches. Are there harder ones to catch? Like slightly faster or slower game pace? On some SMB2 levels certain actions cause one enemy to appear or 2. Example world 4-4 at the beginning right after the fire beam. Depending on your jump you get one or two goomba's. Thought in the light of OCD accuracy I'd ask since I finished the game about 15 times very recently

[hasty river]

Oh snap we gonna get a release soon with all this goodness?

[dry lynx]

New NES and Saturn cores is basically Christmas

[fierce estuary]

I didn't look carefully yet. It's in a totally different part of the core than i've been working on so im going to focus on getting this in good shape to merge first

[fierce estuary]

I don't think much of what I implemented would impact that. NMI and IRQ's are all accurate, processor is all accurate. If anything it would be some kind of sprite 0 behavior with an intentionally corrupted oam, but I don't think any commercial thing would do that in a million years

[gray plaza]

The bug you're talking about with SMB2 has to do with game logic, it has 6 bytes reserved for loading enemies and depending on where the koopa is it doesn't unload that slot for the goomba to appear.

[gray plaza]

PAL through Analog IO seems messed up

One on left is PAL from a September build, one on right is PAL on latest build posted by Kitrinx

Dendy's also got an issue (seems to be broken on HDMI output too)

[steep yew]

Svideo?

[gray plaza]

Component through RGB SCART cable into Keene RGB2C (my crt doesn't like the sync from the SCART cable when trying to use RGB)

All the other RGB related sync stuff is off now so it's essentially just pushing YPbPr through SCART

[steep yew]

lookin'

[hasty river]

I also would love an SNES release with save states and rumble and all that stuff

[steep yew]

Yeah, pal and Dendy are messed up on analog. Dendy is also messed up in hdmi

[mint lichen]

I played thru that Celeste Mario hack on the newer core and didn't notice anything out of place in the mini games either. I'll try a bunch of homebrew stuff tomorrow.

[gray plaza]

I don't know if this is intended or not but PAL PPU on the new versions seems to have created a palette corruption bug with PAL SMB Practice ROM's savestate feature (Note the missing line on the crown)

[main narwhal]

mesen2 is a fantastic emulator for debugging

[fierce estuary]

does it fix itself?

[gray plaza]

Whenever the game modified the palette data yeah it'll fix it, either by reloading the intermediate screen or waiting until something overwrites it. Also I should say this isn't exclusive the the game's intermediate screen

This doesn't happen on the older builds of the core I have

Granted, this could be some weird quirk of PAL hardware that I am unfamiliar with

[livid leaf]

The hack was probably not PAL optimized.

[gray plaza]

This hack is a practice hack for SMB1 (along with 2J and Nippon) and there exists forks for NTSC & PAL optimised builds of SMB1. I don't think the timing for writing the necessary data to RAM for savestates would corrupt the palette this way?

I can't guarantee it though, for all I know there's some weird case with the PAL hardware I am not familiar with similar to 2U Rev 0's weird NMI bug

[fierce estuary]

I didn't think I touched the palette savestates at all

PAL is really shit right now

I have to implement arbitrary ppu cycle alignment to make it okay

[gray plaza]

Oh, this isn’t the MiSTer savestates I should clarify. So sorry. This is Kaname Suite’s (practice hack name) built in Savestate feature

[fierce estuary]

what kind of games is dendy supposed to load

im not worried about it for now then

ill fix the analog though

what IS dendy supposed to look like

also is that a PAL tv?

[steep yew]

Mine supports pal

[granite pivot]

I tried Flying Hero. Supports mouse/spinner when I set the game to Vaus. Tried on AV Famicom and it seems to only support the Arkanoid II Vaus.

[fierce estuary]

here, this should fix pal/dendy analog video

give it a try for me please

[hasty trout]

Yup that fixed PAL and Dendy on my PVM, thanks!

[dry lynx]

PAL lives again!

[gilded idol]

it's interesting that the vaus control for the game isnt mentioned anywhere, even the manual.

[gray plaza]

yeah PAL works now on my CRT

[rare edge]

Pal

Dendy

[steep yew]

Is that on the latest core @rare edge ?

#1091056042667937944 message

[rare edge]

Yes

[gray plaza]

#1091056042667937944 message

Try this one

[steep yew]

Ahh, yeah, sorry, the latest core is the one Lain and I linked

[timber lava]

During some transition screen, I saw some garbage for Doreamon (in attract mode), you will find also the save state. Could someone checked if it is happens on realhardware ?

[dreamy marten]

There's a Dendy option on the nes core?

[timber lava]

yes

[dreamy marten]

What does it do

[timber lava]

You can choose between NTSC, PAL and Dendy on System type option, i think it is the video mode but I don't know the differences

[half pasture]

Doesn’t the Dendy mess with the palette colors?

[mint lichen]

I can confirm this does not happen on AV Fami, it sits on the screen for a while then goes to two more before going to the title screen. I waited several loops and nothing like this happens. And this happened first attract loop on the core for me, so seems like a possible core bug @fierce estuary

[timber lava]

thank you for the check !

[fierce estuary]

ill check it, that's japanese?

[mint lichen]

Yep, the first Doraemon Famicom game

[timber lava]

Doraemon (Japan) (Rev 1)

[fierce estuary]

thank you

[timber lava]

there is also the save state above if needed

[fierce estuary]

it could be just a mapper error on that one

seems like a super simple mapper

[timber lava]

for my rom

[fierce estuary]

Yeah. I'll have to look over the cause when I get home

[glacial turtle]

Any idea how hard it would be to add chr ram support to mapper 26 (for a mega man rom hack) ?

[tropic belfry]

Perhaps widely known about, but this hack is pretty cool. https://www.youtube.com/watch?v=Lg9r9_MKDHU

[jade zenith]

which hack would this be for? it wouldn't happen to be the mega man vrc6 hacks, would it?

[fierce estuary]

there's a new build in #test-builds

there was an ancient addressing issue that impacted the a12 address line

[hasty trout]

@fierce estuary your new build fixes Sword Master!

The sound is now correct

😁

[sharp geode]

oh

[fierce estuary]

really? oops

I mean, of course it does

[fierce estuary]

this corruption happens in mesen too

and nintendulator, all exactly the same

[mint lichen]

Weird, I remembered I actually have the cart, so this is cart on AV Fami https://streamable.com/ynwb2a

[fierce estuary]

oh I believe you, I just believe it's a bug on every emulator that exists

https://forums.nesdev.org/viewtopic.php?t=20994

[hasty trout]

Nesdev to the rescue again, that’s interesting

[fierce estuary]

if you set ram clear to random, it will load different every time

so, not a bug

well, not MY bug

what other game bugs are there, ill add Sword Master to the fixed list

[half pasture]

Does your last test version fix this as well?

[sharp geode]

for me no changes, idk why

[hasty trout]

Oh weird

[sharp geode]

with the lastest test core

[fierce estuary]

is that from save state or fresh?

[hasty trout]

Just to make sure I didn't mishear, this is a recording from the latest MiSTer NES test build. Do you still hear the issue here?

I didn't use a savestate

[sharp geode]

sound is correct

[hasty trout]

thanks 😄 glad I didn't mishear. Weird though

[fierce estuary]

@sharp geode did you try fresh or from save state?