What's new
With MiSTer it’s always going to be about the quality of the core that ultimately dictates accuracy. Sure, games are playable and in many instances better than MAME and in some instances extremely close to real hardware, but this weekly core thing for the Patreon $$$ has got me wondering in all honesty how 100% “accurate” can it be if it’s a one man churning it out... there’s gotta be some shortcuts taken and oversights made?

It is one of the reasons I will keep my PCBs, multis and consoles around and they’ll never be replaced. MiSTer is an essential purchase in my opinion but not a replacement for a real multi by any means.
 
Of course some shortcuts are taken, only a full decapping of each custom chip at gate level could lead to accurate simulation.

I would have plenty of question to Jose (Jotego), knowing well some of the most famous arcade pieces of hardware. Remember when I said MAME was inacurate regarding B chip behaviour and layer masks? Each layer is activated by a single bit (tested on real hardware with every B chip type), so whenever it's not the case in MAME then it's wrong. Now with the work of @WydD and his full B21 reverse-engineering it's been proved at gate level. I wonder how Jotego has handled that case.
To add to the discussion I think MAME devs didn't know which bit was enabling which layer for some chips cause all games using said chips would enable/disable layers 4 & 5 always together for instance. But a small 68k piece of code can give the answer. Not sure why they didn't care to do that.
 
For the games I truly care about, that we play at a competitive level where every little detail matters, I'm sticking with real hardware. In my case that basically means CPS2 and CPS3.

But that said, MiSTer CPS2 is a really big deal release with a lot of eyes on it, so if the accuracy is to the point where the purists can't tell that's good enough for me to run it in another cab. Just won't be replacing my CPS2 multi.

For anything else if the accuracy is in the 90+% range and there's no input lag that's good enough for me. I'm really only picky about fighting games where we really need that frame/timing consistency across setups.
 
Jotego seems the be releasing a lot of stuff, and also have a lot of stuff in the pipeline. Not complaining at all, but wondering if he’s going for quantity over quality at this point to appease his Patreon supporters?
His cores are amazing, but NOT accurate (to the level we here on this forum want). He did spend all last week going back and doing core updates to older cores to your point though.

He's also working with Furtek as chips get de-capped to go back and implement the fixes in his cores based on the information.

So, that being said, yes, he IS cranking out cores like crazy (and not just for Patreons, he releases them free for all after the beta) but he also cares about accuracy as much as possible and updates to do so, its just that, like us, he cannot implement something that no one has access too. That's not a fault in his skill, but rather a lack of knowledge available.

The big thing right now is to support the chip de-capping efforts so all emulation can be improved and potential replacement chips made for real PCBs.

If you want to see what outstanding issues exist for his cores, you can see here (and also submit problems you've found):

https://github.com/jotego
 
From what I can see, the jotego cores are of the more accurate - he knows his stuff, and I'm not quite sure what type of accuracy issues people are seeing. Also keep in mind that inaccuracies/bugs will manifest themselves quite differently than in software emulation.
 
From what I can see, the jotego cores are of the more accurate - he knows his stuff, and I'm not quite sure what type of accuracy issues people are seeing. Also keep in mind that inaccuracies/bugs will manifest themselves quite differently than in software emulation.
I agree, I think he's doing a great job. One example is the stars in the first level of Strider on CPS1 (and Forgotten Worlds). They are not correct in any emulation (including MAME) nor are they correct in his core (as he himself has noted). The reason is that the pattern of the starts in the sky on level 1 are hidden in a chip that no-one has de-capped or has the schematics for (that damn CPS1 A chip). So the the behavior is guessed for emulation, and thus not accurate to the real PCB.

https://github.com/jotego/jtcps1/issues/38

Here is another example where he knows there are inaccuracies in Side Arms (potentially the same code that does the Strider & FW stars), he's trying to hunt down:

https://twitter.com/topapate/status/1311591914035183616?lang=en

These are just two examples, there are many more, however no one is saying MAME or MiSTer are perfect, just that every update gets us closer. Jotego is doing his best, but again, he might be limited based on available information, for now.
 
The cps2 core is in beta so we shouldn’t expect it to be perfect. Dont forget Cps1 core got a big revision after release changing the speed to be much more accurate as well. He might be pumping them out fast but he seems committed to quality even well after cores are released.
 
I will admit that my OCD detector screamed when I fired up Jotego's Bubble Bobble core and saw that the bubbles don't spread out from the middle when you start a game like they do on an OG board.

MiSTer is really cool and deserves much praise (as does Jotego) but it's still definitely not 100% accurate on many, many cores.
 
I will admit that my OCD detector screamed when I fired up Jotego's Bubble Bobble core and saw that the bubbles don't spread out from the middle when you start a game like they do on an OG board.

MiSTer is really cool and deserves much praise (as does Jotego) but it's still definitely not 100% accurate on many, many cores.
I keep looking, and was looking just the other day again, for a list of arcade cores and how "accurate" they are. By some % margin would someone say they represent the authentic feel and experience. I never would have known that issue with the bubbles since I have never played the OG board. Hope vigilance and reporting is kept up for these small nuances. I'm in their discord should I ever get around to getting one of these and find a bug.
 
I keep looking, and was looking just the other day again, for a list of arcade cores and how "accurate" they are. By some % margin would someone say they represent the authentic feel and experience. I never would have known that issue with the bubbles since I have never played the OG board. Hope vigilance and reporting is kept up for these small nuances. I'm in their discord should I ever get around to getting one of these and find a bug.
On some Patreons they will talk about issues and accuracies but to be honest, who's to really give it an accurate %? I mean, until every chip is de-capped, we'll never be at 100%. Even if all those chips are de-capped, who's to say the core designer writes accurate code?

I say just get one, they are cheap enough to mess with (if you go basic and build yourself) and mess around with it. It's fun, but not likely to drive a hardcore PCB user away from the real deal.
 
I'm waiting for the early 90s Midway cores. Once I see a MKII core that slowly deteriorates every time you play it I will be impressed with the accuracy to the original hardware.
 
On some Patreons they will talk about issues and accuracies but to be honest, who's to really give it an accurate %? I mean, until every chip is de-capped, we'll never be at 100%. Even if all those chips are de-capped, who's to say the core designer writes accurate code?

I say just get one, they are cheap enough to mess with (if you go basic and build yourself) and mess around with it. It's fun, but not likely to drive a hardcore PCB user away from the real deal.
It’s so hard to decide what’s “enough”

Is the CPS 2 core stuff awesome...absolutely. Is it accurate like OG hardware? No

would I sell my CPS 2 boards and my Darksoft setup? Absolutely not

when I want to play a quick 20 minutes am I going to fire up my MiSTer and use that vs taking the Darksoft off the shelf and hooking it all up? Yes!

it’s basically “hardware adjacent” to me. Even the Neo Geo core being the most “accurate”...if I want to sit down and play a few hours of KOF I’m doing it on my AES. It just has that little extra “originality” to it.

For me MiSTer excels at the stuff I can’t do easily with real hardware. Am I going to buy a Lynx and try to get RGB out and hook it up to my tv? Hell no...so that core is amazing.

Do I love playing Gameboy and Gameboy color games from my couch on MiSTer? Hell yes.
 
I have played with the MiSTer and Neo Geo core with 3 games and found inaccuracies in the 3 of them, music gets out of tune in lot of games, as if it struggles to maintain a tempo with wrong speed, for example, the music in Art of Fighting in character selection changes pitch all the time and is inconsistent compared to the original and even mame is more accurate in this area... had this problem in KOF98 and SNK VS Capcom, didn't try much more, repoted this in the Github
 
Price of the DE-10 Nano went up by $30 to $170 now...
 

Attachments

  • C2B266E6-70F9-4CEA-A10C-A4BEDC61A025.jpeg
    C2B266E6-70F9-4CEA-A10C-A4BEDC61A025.jpeg
    281.5 KB · Views: 50
I have played with the MiSTer and Neo Geo core with 3 games and found inaccuracies in the 3 of them, music gets out of tune in lot of games, as if it struggles to maintain a tempo with wrong speed, for example, the music in Art of Fighting in character selection changes pitch all the time and is inconsistent compared to the original and even mame is more accurate in this area... had this problem in KOF98 and SNK VS Capcom, didn't try much more, repoted this in the Github

That is one of the known issues with the Neo Geo core. You can see the full list here:
https://github.com/MiSTer-devel/NeoGeo_MiSTer/issues
 
Back
Top