CPS 1.5

[lamarax]

Problem is, Jt "downclocked" his old Street Fighter 1 core in order to accommodate "bad 128MB SDRAM modules" with his new controller, and the thing doesn't even boot up now.

I think we have a "tempest in a teacup" situation, where everything was working fine before, but now we want it to be "perfect" (breaking everything along the way)

[Chris23235]

I can't seem to map coin button. It accepts the mapping but ingame nothing works. 1.5 only CPS1 works fine.

Had the same problem after running the Update All script. Downloaded the mra files from Jotego's repository and they fixed the problem.

[hiddenbyleaves]

I got my ram from https://misterfpga.co.uk a few weeks ago. It plays everything fine except Punisher. The weird thing is that if Punisher is the first thing I boot up after the mister has been off for awhile it works perfect. I can leave it on and play it as many times I want without it crashing. However, if I load another core or even just reset it then it crashes at the beginning of the first level.

[FatSlob71]

So where exactly do we solder this cap to, any diagrams to follow ? I am Busy putting Behringer System 100 modules into my rack, will keep my busy for a while.

[EvilRyu]

I got my ram from https://misterfpga.co.uk a few weeks ago. It plays everything fine except Punisher. The weird thing is that if Punisher is the first thing I boot up after the mister has been off for awhile it works perfect. I can leave it on and play it as many times I want without it crashing. However, if I load another core or even just reset it then it crashes at the beginning of the first level.

This is exactly what happens in my case but with the SF Zero Qsound conversion. Cadillacs and Dinosaurs and Punisher seem to work fine.

My 128MB module is also from Nat.

[dshadoff]

That's interesting, because the FPGA is the part which heats up during use - not so much the RAM.
The FPGA's timing characteristics shift slightly within the operating temperature range, which is generally sent to the compiler as a hint; the default is to accept junction temperatures up to 100C.
But there are several potential scenarios which could cause it to fail timing at higher temperatures; these would be under the control of the programmer - and the compile output should provide visibility to the programmer of whether timing fits or fails, based on the assumptions/constraints in the system.

[lamarax]

I'm pretty clueless when it comes to electronics; one question remains though in my simple mind:

Jt updated his SF1 (not CPS1, nor QSound) core a couple of days ago with his new SDRAM controller, to supposedly accommodate people with "faulty" 128MB kits. Apparently he was unaware of the fact that cutting clock times to half, would render this particular core inoperable. That tells me more than blab about flux quality, electrolytic caps and the like. So who should I trust? Should I wait for someone to manufacture the "correct" module, which I have to buy anew in order to keep up, or should I write off all that stuff and enjoy what I have (including the awesome upcoming Cave 68K core)?

[Alkadian]

I got my ram from https://misterfpga.co.uk a few weeks ago. It plays everything fine except Punisher. The weird thing is that if Punisher is the first thing I boot up after the mister has been off for awhile it works perfect. I can leave it on and play it as many times I want without it crashing. However, if I load another core or even just reset it then it crashes at the beginning of the first level.

This is exactly what happens in my case but with the SF Zero Qsound conversion. Cadillacs and Dinosaurs and Punisher seem to work fine.

My 128MB module is also from Nat.

@EvilRyu,
Just out of curiosity, would you please let me know if Slammasters plays well with your setup. I have got the 128MB ram from Nat as well.
Thanks!

[silentheaven83]

I got the 128 MB from Nat too and it’s very random. Three days ago I couldn’t start the first round of SFZero CPS1.5, yesterday I fought with three opponents.
Before SFZero I played without issue C&D and The Punisher.

It could depend if it’s the first core loaded after the boot or not, I don’t know, but it’s very random.
I'm pretty clueless too when it comes to electronics.

[Alkadian]

I got the 128 MB from Nat too and it’s very random. Three days ago I couldn’t start the first round of SFZero CPS1.5, yesterday I fought with three opponents.
Before SFZero I played without issue C&D and The Punisher.

It could depend if it’s the first core loaded after the boot or not, I don’t know, but it’s very random.
I'm pretty clueless too when it comes to electronics.

Thanks for your feedback. As mentioned in one of my previous posts, only Slammasters plays randomly. All the other games play perfeclty fine.

[silentheaven83]

We could ask if Nat changed something in assembling the boards through time if it does mean something. I don’t know if he’s here on the forum but it’s a very kind person.

I bought my board in September 2020, it’s a black board, v2.5, never had issues with other cores beside CPS1.5.

[dshadoff]

My understanding is that Jotego plans to relax the timing in his code, and get this working on existing modules.

[fernie417]

For whatever it's worth; also having trouble with only Saturday Night Slam Masters and no other game, including the other CPS1.5 games and the Street Fighter Alpha QSound hack. When I try to load SNSM up it won't get past the legal text. I was able to get it running occasionally, but after a few minutes of the Attract Mode running it would glitch out and crash. I have a 128mb ram module I bought in July 2020 from MisterAddons, blue board, v2.5.

[MostroW]

I also had an "illegal instruction" thrown on me with Saturday Night Slam Masters after i won the first round and the game progressed to the intro of the second round.
Other games like Punisher i could play all way through, Cadillacs and Dinosaurs i got up to round 8 before i had to turn it off so i couldn't get to complete it, the whole run was smooth though.

I've assembled my own 128MB module with 2 AS4C32M16SB-6TIN memory chips on the board.

[patamar4]

Saturday Night Slam Masters is some problem with protection or something and has nothing to do with sdram and Jotego already know that problem and he will investigate soon, now case of any error during the game or in attract mode there yes is with sdram.

For those who are not having problem with 128mb sdram and can do the following test:

Load some game of cps1 and as soon as the first image appears load the cadillac and dinossaur and leave in attract mode until you reach the forest stage where the three descend from the car until they pass through a tree, see if they pass this part without giving error

[FatSlob71]

Hopefully he can do something in code to mitigate the issue!

[silentheaven83]

Load some game of cps1 and as soon as the first image appears load the cadillac and dinossaur and leave in attract mode until you reach the forest stage where the three descend from the car until they pass through a tree, see if they pass this part without giving error

This is what I did:

1) Put the plug and turn on the MISTer
2) Arcade -> C&D (World) -> attract mode until the point you mentioned -> no issues
3) Fast Reboot (not the cold one) from the GUI
4) Arcade -> Alternatives -> SFZero CPS1.5 (the one that gave me problems) -> attract mode for 2 minutes -> no issues
5) Started the game with Ryu vs Birdie, Ryu vs Rose, Ryu vs Adon -> no issues
6) Fast Reboot -> C&D (World) again -> Attract mode for 10 minutes -> no issues
7) Fast Reboot -> SFZero CPS1.5 again -> Attract mode for like 2 minutes -> ADDRESS ERROR - ADDRESS: 00023B64

It's totally random, few days ago I could not start a single round of SFZero. I only think that it doesn't happen right after you boot the MISTer.

BTW it seems to run really hot (I got only an heatsink and no case right now), is this a hint that Jotego is "pushing the limits"?

[PikWik]

like the majority of you,
i have been waiting for the CPS2 core and watching/refreshing jotego's twitter/patreon page everyday looking for progress

i am doubtful CPS2 will be released tomorrow (but a christmas miracle could always happen).
i would guess at this point, we will see CPS2 at the end of January-ish.

-------------------

i have been lucky and not run into any memory issues with my 128 module, and i have not had any address errors when i run CPS1.5 games.
i am using a v2.5 128 RAM module bought ~12 months ago from a seller on the atari forums.

i know @patamar4 has done the 10uf cap mod mentioned by jotego and would be interested if they can run the test suggested above, after the mister has been running for a couple hours.
the cap mod is very easy to do if you have a soldering iron and a spare 10uf cap laying around, so im glad it will not be too big of a deal for people that are getting the errors with their modules.

perhaps jotego can create a workaround in his programming for his CPS1.5 core and his CPS2 core,
but i would guess jotego doesn't want to remake his RAM controller and re-write a ton of code to facilitate for 128 RAM modules that are supposed to function how he was expecting. jotego has said he has ten different 128 modules that hes tested with his cores and 2 of them have voltage variance in their capacitance that is setting off the memory errors.

Tl5E6gJ.png (52.73 KiB) Viewed 11901 times

he has also shown that when he added a 10uf capacitor in parallel to another 10uf cap on a "faulty" 128 module, this corrected the voltage variance and he was no longer getting the errors from before the mod.

flQh0JH.jpg (1.48 MiB) Viewed 11901 times

v2.4/v2.5 cap layout
top row: 10uf, 0.1uf, 10uf, 0.1uf
bottom row: 10uf, 1uf, 1uf, 10uf, 10uf, 1uf, 1uf, 10uf

(please note the polarity of the 10uf cap you're using before soldering)

[dshadoff]

perhaps jotego can create a workaround in his programming for his CPS1.5 core and his CPS2 core,
but i would guess jotego doesnt want to remake his RAM controller and re-write a ton of code to facilitate for 128 RAM modules that are supposed to function how he was expecting. jotego has said he has ten different 128 modules that hes tested with his cores and 2 of them have voltage variance in their capacitance that is setting off the memory errors.


he has also shown that when he added a 10uf capacitor in parallel to another 10uf cap on a "faulty" 128 module, this corrected the voltage variance and he was no longer getting the errors from before the mod.


(please note the polarity of the 10uf cap youre using before soldering)

I wouldn't characterize it like that - the modules were never designed or produced with that access method in mind (they were designed to support certain speeds with the existing SDRAM controller code), so the phrase "supposed to function how he was expecting" does not ring true.
He can relax the timings, and I understand that he intends to do so, so that the cores will work with existing hardware.

It should be noted that the following day, he wasn't able to elicit the error on any of the modules with the RBF which had shown them prior (based on conversation from his discord), so consider this an ongoing area of investigation.

Everybody should just calm down a bit, and refrain using language implying blame or fault.

[PikWik]

oh for sure!
i dont mean to put blame or point fingers at all.
im just compiling things ive read/seen about this topic.

i am hopeful jotego will work some magic with CPS1.5/CPS2 (downclock to 64mhz and simultaneously increase data throughput) and all will be well.
at that point, no one will need to mod anything and people that are selling 128 RAM modules can rest easy knowing they dont have to re-purchase 30 new pieces for their webstore.

[PikWik]

the hype for CPS2 is real and will mark a moment in emulation history, where arguably, the greatest era of capcom fighting games will be in FPGA

Apd7CRW.gif (1.62 MiB) Viewed 10054 times

[grogre]

I think there are lot of things that could be involved in the arcade folder in Mister, sometimes things don't work, this part of mister is a bit more complicated than the console part and it can explain some difficulties.
I consider that a part of the problems is not due to ram but in mistakes with the arcade directory and the way to use it, i've made this experience, i thougt it was my ram modules but in fact a simple "blind" mra file that couldn't find the sfz 1.5 rom properly.

[lamarax]

Happy Xmas everybody!

[RetroP]

Had the same problem after running the Update All script. Downloaded the mra files from Jotego's repository and they fixed the problem.

Worked a treat!

Thank you

[Vespa]

I have128MB ram module from https://misterfpga.co.uk and the latest update now makes Punisher and C&D both pop-up with address errors where before they worked fine :/ I think Jotego said he's done messing with it too fuuuuuu

[suverman]

I have128MB ram module from https://misterfpga.co.uk and the latest update now makes Punisher and C&D both pop-up with address errors where before they worked fine :/ I think Jotego said he's done messing with it too fuuuuuu

Send it back and ask for a replacement. Ricardo, Pork and Zerohimself have already said they will replace if people have issues with their RAMs.

[Nat]

I have 128MB ram module from https://misterfpga.co.uk

Sorry to hear your now having issues with Jotego's latest build, if you are one of my customers with a v2.5 128MB SDRAM module and having issues, just contact me. I will rework or exchange your SDRAM module.

[Vespa]

I have 128MB ram module from https://misterfpga.co.uk

Sorry to hear your now having issues with Jotego's latest build, if you are one of my customers with a v2.5 128MB SDRAM module and having issues, just contact me. I will rework or exchange your SDRAM module.

Hey Nat, good to see you here! I just wanted to check if I was on the latest build (appears I am) and redo the RAM test. So I get no Red screen and there's no sound but the OSD led will flash once every 30 seconds, not sure if this counts as 'blinking' and therefore a fail?

Do you think it's best to wait for the CPS2 beta core to drop before exchanging the RAM? Save faffing around again if there's some unforeseen issue with that.

[lamarax]

There's already a new v3.0 board in production (a cooperation between Sorgelig and Ricardo), but frankly, I don't believe it's needed for the time being and with what we have.

Jotego should examine his proprietary SDRAM controller more closely, instead of issuing "tests" left and right.

[suverman]

There's already a new v3.0 board in production (a cooperation between Sorgelig and Ricardo), but frankly, I don't believe it's needed for the time being and with what we have.

Jotego should examine his proprietary SDRAM controller more closely, instead of issuing "tests" left and right.

I believe you are wrong. Also JT has mentioned he will go back to core development from this week.