#Sony PlayStation

1 messages · Page 8 of 1

visual depot
#

It needs to be named boot.rom

simple otter
#

I have 3 as per mister's recommendation

#

502224B6 - boot.rom
502224B6 - boot1.rom
318178BF - boot2.rom

visual depot
#

Ok so neverminnd

simple otter
visual depot
#

I didn't think converting to CHD would change anything, but CHD works fine. Direct BIN/CUE doesn't

#

I've reproduced the issue with the single image BIN/CUE

#

@tropic stratus Can you take a look at Twisted Metal 4 (OG release)

#

Not sure but it seems it has some music issues, but converting to CHD fixes it.... I know CHD transcodes the audio

simple otter
#

I started out doing chd but changed my mind when I reconsidered that hashes and the integrity of the media were important to me

visual depot
#

nah CHD actually is lossless

#

But it losslessly reencodes the audio to FLAC

#

which saves a LOT of space

#

there's no loss of audio frequencies

#

FLAC is a lossless compression, just actually can detect null audio frequencies. Archive formats can't do that

oblique frost
#

and as a result, can be unchded back into your original redump worthy bin/cue

visual depot
#

Fyi I did use the wrong word. It doesn't resample. Just transcodes to flac to my knowledge.

simple otter
#

Can confirm that my .img converted to .chd functions exactly as expected

tropic stratus
#

Twisted Metal 4 is a multitrack game, you cannot play it with single track bin/img, only chd

#

the reason is that the single track bin code in mister main is broken. It was implemented halfway by Caldor(?), then left without finishing it

#

it should probably either be finished or the HPS could warn you that the cue contains multiple tracks in a single file and that is unsupported

#

i honestly won't touch it. Duckstation contains a hint about single bin - multi track images and does a lot of guessing to try to get it right. But I don't like guessing, it will only lead to random issues

#

So officially, multitrack disk support is only there for multi bin/cue. CHDs seems to be fine as far as i know. But multitrack-singlebin is known to be broken

visual depot
#

Oh I didn't get that was the reason why. My mistake.

#

I thought multitrack bins didn't work either though. @tropic stratus

#

People here have said you need single bin.

wanton crown
#

If more people could try to reproduce this issue (so far I'm the only one on two different brand new mister fpga setups)

ivory verge
visual depot
#

LOL

#

@simple otter have you tried multibins like the og rip?

#

I will try myself tomorrow

ivory verge
#

If the game has CDA music, each track must be separate ( bin)

ivory verge
#

Just kidding ofc

wanton crown
#

Ahah I'm just worried if different de 10 nano revisions could cause issues... terrasic's revision document was not update since 2019 but that was before covid and component shortage...

#

I love the mister project mainly for preservation but if we start getting glitches like this because of revisions it's a bit annoying

visual depot
#

The core is new mate. Cores need work too just like emulators. It's not possible to be perfect with such a new core 😛

wanton crown
#

The thing is I'm the only one so far with this issue. Simular but more random issue also occurred in cps2 core.

visual depot
#

Things can happen as a result of a revision without any understanding that it would cause an issue.... It's not done out of oversight or hurriedness. This isn't a simple language.

wanton crown
#

But I was also the only one who tried with a fpga bought at least 4 months ago

visual depot
#

Well I'm about to head to bed. Someone will test it I'm sure but it's like 4:30AM in the east coast and 1:30AM on the west coast of the US.. Might not see as much activity until a lot later.

wanton crown
#

I'm not critizing anyone or anything, just that if developers start to have to deal with different de 10 nano boards revisions that have different ddr3 speed or whatever, it complicates everything

visual depot
#

The de10 nano has a standard design. It doesn't change

wanton crown
#

At least 4 people have tried reproducing to no avail so far.
I did on two different mister fpga setups, changed sd cards also

visual depot
#

Lot of people with new de10's without issue.

wanton crown
#

Given how little and finicky the issues we are talking about are, I wouldnt be surprised

#

As most people with new misters dont play games in japanese and dont play sf2x on cps2

visual depot
#

Now if you used the same sdram I'd start having questions about that. But let's see if it's reproducible. Sometimes it's something requiring a savestate with specific factors to reproduce, which can help troubleshoot

#

That's why we ask for savestate

ivory verge
#

We checked the save , also started the game from the beginning, with no result

wanton crown
#

If I can reproduce 100% this issue on two brand new misters bought from terassic, from nothing but a fresh mister on the sd card that came with the de10 nano, the update script and copy pasting redump verified bin/cue of the game, I dont see how it can be setup specific

visual depot
#

Did you get new addons too?

#

Terasic doesn't make the other stuff

wanton crown
#

Yeq i think Kuba used the same sd ram as I did , mem test showed nothi g

ivory verge
#

We can't reproduce it, I've checked every possible option in core

wanton crown
visual depot
#

Huh? Do you live nearby kuba to give him the sdram stick you have?

wanton crown
#

Reproduced with two different 2.9 sd ram stick

#

That both pass memtest at 150mhz

#

Also just using the usual usb board

visual depot
#

And does it happen on a different tv?

wanton crown
#

Happens on crt and hdmi screen

#

Doesnt happen on real hardware at all.

#

I have my ps1 hooked and tried right away before reporting.

tropic stratus
#

will try it now

wanton crown
#

I can send you my whole mister unit if you want to try

#

Germany is just nearby

tropic stratus
#

if the issue really started for you with 10 21, it's most likely sdram related

wanton crown
#

Interesting

tropic stratus
#

the other changes in this build are very unlikely to be a reason

visual depot
#

We don't actually own a shop.. people in the community do. There's like 4+ shops we generally see that people buy from. Which one was it?

wanton crown
#

Is there any recommendation besides using those specific sd ram stick ? Kuba reproducded with exactly the same. Unless I got 2 bad sticks... that happen to still pass memtest

tropic stratus
#

let us try a few things, but first i want to test myself

wanton crown
visual depot
#

Alrighty

wanton crown
tropic stratus
#

anything i have to do after loading the savestate?

wanton crown
#

No. The glitch occurs right away for me

tropic stratus
#

but it's not in the exact frame that is loaded...

#

so only every second frame, therefore the flicker

visual depot
#

The fact that it happens in CRT and HDMI boggles me... But it looks like something that could be hidden by the interlacing. Not sure.

tropic stratus
#

please activate the vramviewer : l3+r3+select when not in osd, then go in osd -> debug ->ddr3 framebuffer + vramviewer

#

only works with hdmi

wanton crown
#

Looking into it, let me get out of bdd first LOL

tropic stratus
#

sorry 🙂

wanton crown
#

Bed *

#

Dont be sorry Im thankful

tropic stratus
#

would be curious if it only happens in the upper/lower framebuffer consistently

visual depot
#

Hi thankful I'm jerk

#

😂

tropic stratus
#

in this situation, i would play around with turbo and see if it changes

#

hmm, turbo high already breaks the game

ivory verge
#

Yes

#

Works to mid

wanton crown
#

Hm I think I changed turbo and it didnt change anything

#

Reporting in 5 min

ivory verge
tropic stratus
#

that is a different room the savestate i'm trying? i'm confused

ivory verge
#

Yes, I'm just checking other locations

#

Sorry

tropic stratus
#

if the issue is already there instantly after loading the savestate and it still behaves different, this means the data must be in the sdram already

#

the flickering is like either a additional triangle is drawn, or it's some effect when clearing the image. In either case it doesn't make sense with a faulty sdram, as this would rather mess up the GPU completly or randomly

wanton crown
tropic stratus
#

so it's always in the upper image?

wanton crown
#

Yea

#

quality is shitty sorry, but indeed the triangle doesn't flicker and just stays there on the upper part

#

I've not seen such behavior in any other game on the core tbh, everything has always played fine.

#

btw, mouse hovering the elements that interactable makes the flickering go away totally

tropic stratus
#

to debug this further it would require simulating the frame, then comparing the single gpu commands from signaltap against the simulation, then dig where the data comes from in ram, see how it was written there, ...

#

one trick we might have is to crash the cpu, so the ram content is 100% static and compare such a crashed savestate against your mister

#

scratch that, the image would not get rendered....

empty yacht
#

For it being a power supply issue would be too convenient and reproducible, right?

tropic stratus
#

yes

wanton crown
#

I'm using a very good supper supplies on both mister fpga also

tropic stratus
#

i'm running a simulation now, trying to find if this triangle should be even there and if yes, at which point in the image rendering

tacit fox
#

Tried swapping RAM modules ?

wanton crown
#

I only have those two 128 mb sd ram sticks

tacit fox
#

I mean, you have two misters right?

wanton crown
#

yes, both show the behavior

tacit fox
#

Try to exchange the ram modules

#

Oh gotcha

tropic stratus
#

same effect when turning of textures?

#

this will alter gpu timing a lot

wanton crown
#

yea same thing when turning textures off

tacit fox
#

Send your de10 nano to Roberto for debugging

wanton crown
#

If he wants, I can provide it to him no problem.

#

Now, I'm pretty sure he's busy enough and since I'm the only one impacted...

tropic stratus
#

well yes ... honestly, this seems a bit much work for a minor thing in one game only happening for 1 user 😕

tacit fox
#

@wanton crown can you please send savestate and steps to reproduce? I wanna give it a try

tropic stratus
#

#1046940919607345272 message

#

savestate is here, you just need to load it

tacit fox
#

Thank you

wanton crown
#

game is Clock Tower Ghost Head, japanese version

tropic stratus
#

you don't even need the game, no cd access at this spot 😅

tacit fox
#

😱

ivory verge
#

Does behave the same way? Maybe it's worth checking

tropic stratus
#

the game does some strange quad drawing as the very last action before the image is done....a stretched untextured black quad with very strange coordinates(-23/27, 3/162,-157/26,-116/168)

#

but this still doesn't explain the effect

#

only if 1 of these dots would be more inside the image

ivory verge
#

This is said to be happening in other locations as well,

#

@wanton crown It always looks the same?

wanton crown
#

yes, in almost every room I can see it pop. In this one and particular spot it was just more easy to see

#

hmm the flickers are always in the form of triangles, and they are usually on the left side of the screen

#

they are no always at the same place

tacit fox
#

No flickering artifacts here

#

What video mode and sync mode are you using @wanton crown ?

full nimbus
#

Is it a pure line from the Mister to the Display Device or do you have switches capture cards between the Mister and Display Device

ivory verge
wanton crown
#

Stock mister.ini with just composite_sync set to 1, but Kuba had no issues with it

full nimbus
#

and is it a direct connection from the mister to the display

#

Ie No added crap

wanton crown
#

Yes

empty yacht
#

Tried stock mister+ram only?

wanton crown
#

Yes just did right now

#

Forgot who asked but here is a pic of the mister

#

I mean de 10 nano board

empty yacht
#

different room in the house?

wanton crown
#

@scenic reef here is a pic of the de 10 nano board

20230115_112234.jpg
tacit fox
wanton crown
#

Some rooms show the same behavior unfortunately

tropic stratus
#

the problem is already in the framebuffer in vram, so video mode or output device is don't care

#

cd image also

ivory verge
tropic stratus
#

so it boils down to core timing or internal fpga damage(very unlikely) or sdram/ddr3 issue

ivory verge
#

These settings are not ok

tropic stratus
#

what do these switches even do?

ivory verge
#

Should looks like this

wanton crown
#

is it normal that if I plug the sd ram on the other port, it is not recognized by the board 🤔 ?

full nimbus
#

set up the FPGA

tacit fox
#

Good catch @ivory verge

tropic stratus
wanton crown
#

ah I see thanks

tacit fox
ivory verge
#

Set dip correctly, and try again

full nimbus
#

the dips muist match what Kuba posted

wanton crown
#

trying right now !

empty yacht
#

They do this

image.png
#

some HPS stuff

ivory verge
wanton crown
#

they are actually set like in that picture just above.

full nimbus
#

in your photo they dont look like that

wanton crown
#

yea because it's a side view, so unfortunately everything was ok ...

full nimbus
ivory verge
#

Ok then maybe it looks different in the picture, back to the beginning 🙂

wanton crown
full nimbus
#

always worth switching them back and forth

#

to the correct positions

tacit fox
#

Great optical illusion

wanton crown
#

1 UP
2 DOWN
3 UP
4 DOWN
5 UP
6 UP

ivory verge
#

Yeah looks good

wanton crown
#

yea lol

ivory verge
#

I apologize for the confusion

wanton crown
#

no worries. Too bad I thought we finally had a clue 😂

tacit fox
#

Any chances you having a 32mb ram module lying around?

wanton crown
#

unfortunately I don't 😦

#

PSX core only works with 128mb right ?

ivory verge
#

Nope

tacit fox
#

Nope

ivory verge
#

Even with 4mb haha

tacit fox
#

32mb are very cheap

wanton crown
tacit fox
#

Maybe it’s worth it giving it a try?

wanton crown
#

wait so what's the purpose of 128mb ?

full nimbus
#

Neo Geo

#

and CPS2

#

currently

wanton crown
#

those only work with 128 mb ? interesting

full nimbus
#

NeoGeo has very big games

wanton crown
#

I thought people played cps2 with 32mb

full nimbus
#

CPS2 > 64 mb

#

on some

wanton crown
#

I've played a lot of metal slug 3 (which requires 128 mb If I understand) on the neo geo core and never had a single issue

#

I think Kuba tested this very same ram as well and he still couldn't reproduce 😦

full nimbus
#

mr Fpga.co.uk does very good work its highly unlikely to be his ram

#

I have ram from him and dont have your issues

ivory verge
#

I checked on several versions 2.2, 2.4, 2.9, even the old 1.1 with modification

full nimbus
#

From Jotego GitHub about CPS 2

#

All CPS1 and CPS1.5 operate correctly on a 32MB SDRAM system. Some CPS2 games may require a 64MB module. Those games are listed as having a GFX ROM of 16MB or more below. Please check the CPS2 ROM size section.

ivory verge
#

@wanton crown where you live?

wanton crown
#

well it all seems to come down to the de 10 nano board

#

France

ivory verge
#

If you want I can send you sdram for testing

wanton crown
#

I bought the two sd ram some days apart, seems highly unlikely that I could end up with a bad batch

#

don't know if it's worth the hassle 🙂

full nimbus
#

They pass the Ram test so unlikely to be the culprit

tacit fox
#

32MB ram modules are more tolerant to higher timings

#

Anyone remember FF bug on SNES core? 🙃

#

I was one of the affected

wanton crown
full nimbus
#

you may want to hit B on a conroller when running the test to switch to the Other Ram chip

full nimbus
#

Keyboard

Up - increase frequency
Down - decrease frequency
Enter - reset the test
C - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.

Controls (gamepad)

Up - increase frequency
Down - decrease frequency
Start - reset the test
B - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.
wanton crown
#

ah, so I only tested a single chip ?

full nimbus
#

so it seems you can switch between the chips on ram test of 128 mb

#

from

tropic stratus
#

if 10 20 works and 10 21 does not: the only thing changed is how data is fetched from sdram for DMA: the sdram is no longer read in address set + long burst, but instead bank is kept open for the whole DMA transfer, unless a bank is switched. The only other effect of timing being slightly different doesn't explain it, since this would hit everyone else, too

wanton crown
#

:/

wanton crown
tropic stratus
#

on the other hand sdram seems very unlikely, seeing how reproducible it is

full nimbus
#

The docs may be out of date

wanton crown
#

huh on first chip I'm getting errors at 150mhz 🤔

full nimbus
#

150 is very fast

tacit fox
#

It’s normal

full nimbus
#

140 is fine

wanton crown
#

hm ok, going to 140mhz then

ivory verge
#

Please try us version

tacit fox
#

There’s also a jap demo

wanton crown
#

trying us version

#

so memtest passed on both chips with no issues at 140mhz

#

tried the US version of the game, the same bug occurs

ivory verge
#

Ok thanks for testing

wanton crown
#

are you also using a heatsink ?

ivory verge
#

Yes

lost summit
#

Is your PSU is good enough ?

wanton crown
abstract drift
#

did you say you're using the sd card that came with the de10?

ivory verge
#

Yeah is very good psu

wanton crown
#

it's by using that old card (mister fpga ready) that I realized it was the PSX core version triggering the issue for me

timber veldt
#

Can someone link me to the problem @wanton crown has?

wanton crown
#

it would be cool if someone who bought a de 10 nano after me and a little before could try to reproduce

timber veldt
#

TL;DR

abstract drift
#

@wooden rose when did you buy your most recent de10?

wanton crown
# timber veldt Can someone link me to the problem <@187687726420590592> has?

it basically starts here : #1046940919607345272 message
there's a savestate just below
On the game Clock Tower - Ghost Head (japanese version, US version is simply called Clock Tower 2), I have black triangle artifacts, that I double checked on real hardware to make sure it was not the game. People have tried the savestates themselves and nobody can reproduce the problem, but I can on two separate mister fpga setups (that both have a de 10 nano board bought less than 4 months ago)

timber veldt
#

Ok

#

Where did you get that rom?

#

Dump I mean

wanton crown
#

redump, everyone was using the same one (we double checked the hash)

timber veldt
#

Ok

#

CHD?

wanton crown
#

and we went as far as checking the hash directly on the mister

ivory verge
#

Bin

wanton crown
#

no, bin/cue

timber veldt
#

Try CHD

#

And say some holy words

#

Holy water is optional

wanton crown
#

of course using latest release of the core, and I can't reproduce the issue with anything (strictly) before the 2022.10.21 unstable release
but again, that's only me

timber veldt
#

Since you tried other SD cards, deleting cfg files will probably not help

wanton crown
#

yea I also tried renaming the config folder

#

chd changes nothing

tropic stratus
#

you don't need the game, just rename the savestate to some game you have and load that other game 🙃

#

of course it's unplayable, you cannot change rooms, but for the bug it doesn't matter

wanton crown
#

ah well I think I've tried everything I could 🥲 I will continue my clock tower frenzy on real hardware, and hopefully someone with a more recent board could try a reproduction

tropic stratus
#

unfortunatly we all ran out of ideas 😦

wanton crown
#

on a final note, here's how it occurs during normal gameplay

abstract drift
#

what controller are you using?

tacit fox
#

Do you use a fan?

wanton crown
#

yes with the IO board, and it changes nothing

wanton crown
abstract drift
#

you don't have any cheats on or anything do you?

wanton crown
#

no always disabled

tacit fox
#

No MGL files right?

wanton crown
#

no, I only use bin/cue files

#

all other games work perfectly fine, this is the first bug I encounter on the psx core

rugged loom
#

Hi chat 1st time writing here. Just to try to help w this isue.

#

I think noone has asked about BIOS file. Check ur using a jap BIOS or the recomended one or chechsum.

#

gl

full nimbus
#

there are 3 bios files you need one for each region

#

if you use Update all with the Bios Getter script enabled it will get the correct ones and name them correctly

wanton crown
#

I used update all when installing both sd card :/

full nimbus
#

Update all isnt your issue though you just have an unique set up

#

boot.rom => US BIOS
boot1.rom => JP BIOS
boot2.rom => EU BIOS

wanton crown
#

just did a fresh install on a new sd card, just ran the update script, setup the bios by myself this time, and still shows the behavior.
At this point it seems to rule out a different revision or defective de 10 nano, I don't know

full nimbus
#

There is nothing left we can suggest. We have tried to reproduce the issue and no one else has it. Please dont derail other peoples questions with a restatement of your issues

eternal vapor
#

Whose question did he derail? 🤔

full nimbus
#

Hectics

eternal vapor
#

No, hectic was suggesting that he check his bios setup. He was only responding to the suggestion

full nimbus
#

ah I see my Apologies then

#

I thought he was asking about which bios to use

eternal vapor
#

Yeah, I can see the misunderstanding. Just wanted to clear it up

full nimbus
#

Thank you as I say my apologies for misreading the statements

gilded trellis
#

I tried but savestate doesn't work for me

#

Anyway i'm not saying any artifact playing

eternal vapor
#

Just tried it on my setup too (bought December 2021) - couldn't reproduce it.

ivory verge
#

They need to match

gilded trellis
#

Yes..but doesn't work

#

First i saved state for match filename

#

Only works my own save files

#

Maybe crc isn't matching

wanton crown
#

@gilded trellis when did you bought your de 10 nano ?

tacit fox
#

@wanton crown did you try with another game image, like a CHD one?

wanton crown
#

yea, I've done everything I and we could.
I'll just wait for someone with a recently bought fpga to try to reproduce at this point

gilded trellis
#

I bought this september

#

(i reserved it 1 year ago)

#

I tested on a stable psx core

wanton crown
#

didn't know terrasic allowed reservations 🤔
so you don't get artifacts like those I've shown at all ?

gilded trellis
#

nops..i only played 5 minuts..

wanton crown
#

well thanks for trying

ivory verge
#

I understand that it is very annoying when you have a issue and nobody can reproduce it

tacit fox
#

Will purchase a new de10 nano within the next 2 months

#

Will test this for sure

wanton crown
#

I noticed the DDR3 chips reference on the de 10 nano board has changed a lot over the years, I see some ISSI 1532, ISSI 160B, ISSI 1603, ...
they do have different specs (max clock frequeny can range from 666MHz up to 800MHz) but I have no idea if they could matter here

wanton crown
#

if someone who can't reproduce the issue can communicate his de10 nano DDR3 reference, I could buy the same chips and desolder mines off the de 10 nano board and solder the new ones just to confirm

wooden rose
wanton crown
#

my savior

#

I'll be going to workplace on tuesday where I have a third mister (yes I know...), I'll try on that one as well tuesday in the evening

wooden rose
#

you have tested your memory right?

hard laurel
#

Best thing to do. I think he is HORRIFCLY bad YouTuber.... Poor content, doesn't know what he is talking about half the time. Should just stick to playing Metal Music with his friends in their band.

wanton crown
ivory verge
#

Each game must have its own folder

full nimbus
#

and there is nothing wrong with CHD. Its an entirely lossless reversible format

simple otter
#

MiSTer touts CUE+BIN and CHD support, no caveat or asterisk listed

visual depot
full nimbus
#

its on the Github page for the Memory test core

dry drift
#

Pins lack context so they really need a label describing what’s in it

full nimbus
#

changed the original post

dry drift
#

I still don’t get it 🙂

#

It’s for memtest? Not for psx?

full nimbus
#

yes

dry drift
#

Label it as such within the same post

full nimbus
#

It is

visual depot
#

Yeah I know this isn't the ideal channel.

#

But it's really useful info.

dry drift
#

It isn’t

visual depot
#

It's memtest.

full nimbus
#

I edited it to say Memtest

visual depot
#

The same post that's linked.

full nimbus
#

Keyboard and Controller shortcuts for the Memory Tester Core (edited)
[10:44]
Keyboard

Up - increase frequency
Down - decrease frequency
Enter - reset the test
C - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.

Controls (gamepad)

Up - increase frequency
Down - decrease frequency
Start - reset the test
B - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased. (edited)
visual depot
#

You edited the one above it.

dry drift
#

You didn’t

#

I hate these arguments

#

Memtest
Keyboard

Up - increase frequency
Down - decrease frequency
Enter - reset the test
C - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.

Controls (gamepad)

Up - increase frequency
Down - decrease frequency
Start - reset the test
B - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.
dry drift
full nimbus
#

thats the edited post above

visual depot
#

You you want everything in the same post, with a label at the top of it.

#

You edited the post prior to it so it wouldn't show in the pins 😛

#

I'm not trying to be a prick. Sorry.

dry drift
#

I should just copy paste to start with no one knows how to edit a comment

visual depot
#

I was gonna say if you were going to do that just post it in #help

#

Lol

#

Or #mister-hardware

dry drift
#

I’m not sure why it’s in psx but unlabeled in psx really makes people think it’s about psx

visual depot
#

It was fitting the conversation but the info isn't pinned anywhere. Which is why I requested it. Somewhere is better than nowhere.

#

I've tried to find that info many times and never can. I always have to ask.

wooden rose
#

isnt it on birdy's wiki?

#

it is not

rugged loom
wooden rose
#

i think we should just have a general troubleshooting page with "think you've found an issue? first, use the memtest core to check if your hardware is functioning correctly - then, identify the reproduction steps"

mystic violet
#
  • redump bug reports only
summer sage
#

why are you making people memtest? has anyone's memory ever failed to pass it and that was the root of the problem?

#

it seems uncommon

scenic reef
#

More commonly associated with arcade cores.

#

But rare

#

You are confused. Multi track single bin is not anything that matches redump verified dumps. People should use the redump ones instead of merged bins.

#

Chd multitrack works fine afaict

summer sage
#

on nes the #1 problem is that people have bad roms and the #2 problem is that people mistake original behavior for a bug

#

im not actually sure there's been a real bug reported in years

scenic reef
#

On NES? Yeah it's probably been over a year at least lol

summer sage
#

several bug reports, but none of them real

#

anyway it'll probably be similar here

#

cd's are just as bad as nes headers in terms of bad rips and things

scenic reef
#

If it isn't redump then I wouldn't trust it. Merged bins are known to be iffy on lots of CD system emulators.

summer sage
#

im not even sure how they work, where do they store the index

scenic reef
#

It's just the cue pointing to each index within the same file instead.

summer sage
#

oh

scenic reef
#

They just concatenate them together

summer sage
#

I mean in theory that's okay but in practice it's harder to find out if they screwed up doing it

scenic reef
#

Exactly

summer sage
#

i'm a fan of chds now anyway

scenic reef
#

They work fine for me.

summer sage
#

nice tidy compressed single file bundle

#

lightly seasoned with metadata and drizzled with data delivery optimizations

scenic reef
#

Just needs a couple upgrades to the format in the future. There are faster and more efficient compression methods than it uses and the subchannel data apparently isn't carried with it.

summer sage
#

the longer they wait the harder it will be to change

#

as everything integrates support

#

subchannel data is good, but imho they shouldn't ever touch compression

#

the die is cast there

#

having a situation where a device can handle some but not all chd files with no way to tell the difference is not worth it

visual depot
# summer sage I mean in theory that's okay but in practice it's harder to find out if they scr...

Using proper tools, it's generally the correct data. The conversions are usually issue-less (hell it's the main way to convert them to PSP or use them with Android emulators and also the PS2 POPS emulator), as long as you use the correct tools. But there's mainly 2, I think, and both do it right as long as everything was good dumps to begin with: CDMage and MPF (the official redump app: http://wiki.redump.org/index.php?title=Disc_Dumping_Guide_(MPF))

The likelihood of the single bin/cue being bad is non-existent if the original bins were correct, as the cue points to the tracks in the same way it would point to the separate tracks. It just references them in-line in one file instead of separate files.

The issue Robert has explained is the MiSTer main's implementation of singletrack bin code is broken

#

#1046940919607345272 message

visual depot
#

-Robert

#

Surprising to me that this was never finished. I'd never known that was a MiSTer main limitation

summer sage
#

eh that's just sloppy then

#

I would have just assumed it worked with the same code as regular bin cues

visual depot
#

the .cue layout is slightly different, so it would require a slight change to how File I/O works

#

but the bit-for-bit data is the same

#

To give you an idea

#

there's always only one data track, which is always the first bin to my knowledge

#

normally it just lists them all with 0 index, because it can just set it to index 0 of every file

#

when it's one file, it just has the specific index (which is always in time for these) for offsets to each track

#

my guess is the code to load specific "indexes" this way was non-existent

#

I assume the broken music in games is because anything in the data track probably works just fine with the joined bin/cue..... because it's always track 1, and the offsets are 100% the same. But if you try to load track 3, 4, etc for music, it probably is trying to access the tracks assuming they are attached to the first bin file. Unfortunately, since the reference is unreachable, I assume the game just gets null data returned or the last bits in memory possibly

#

so it could be anything from looped audio junk or even nonsense bits that produce ear-piercing sounds

#

These are all educated guesses by the way, but I'm pretty certain this is what it comes down to

#

CHD isn't a zip, contrary to popular belief. It has its own standard reference file for the data structure.... so it bypasses the cue/bin limitation because the resulting chd of a multi-bin is the same of a single bin. In fact, if you un-chd a file, it only ever outputs a single bin/cue because the data is technically one track after the other, bit for bit, on a CD

#

Honestly, think of it this way: Idk if you've ever seen a single-flac/cue music CD rip. But the same thing can apply. The tracks are just referenced with their offsets instead of separate files. Any CD, music or data, is all one big bitstream, realistically, and it's indexed by time offsets. So both single image and multi-track rips are 100% accurately representing the data, minus removed gaps, which are often accounted for in the cue file's gap index if you are using the right software to create the rip and cue file.

forest trench
#

There's a bit of game preservation history where audio gaps were quite a contentious issue (albeit for a different disc format)
See the years of drama between Redump & TOSEC re: Dreamcast GD-ROM backup standards

tropic stratus
#

the problem with (merged)single-bin multi-track is the guesswork for the pregap. Does it have 2 second pregap or not? Some images have them build in, some force it over the cue. For multi track it's easy, as there is a standard. For merged single bin there is no real standard and that's why people get 2 seconds cut off, first 2 seconds looping forever, ...

#

just look at the comment duckstation has for merged bins:

#

"Two seconds pregap for track 1 is assumed if not specified.
Some people have broken (older) dumps where a two second pregap was implicit but not specified in the cuesheet.
The problem is we can't tell between a missing implicit two second pregap and a zero second pregap. Most of
these seem to be a single bin file for all tracks. So if this is the case, we add the two seconds in if it's
not specified. If this is an audio CD (likely when track 1 is not data), we don't add these pregaps, and rely
on the cuesheet. If we did add them, it causes issues in some games (e.g. Dancing Stage featuring DREAMS COME
TRUE)."

#

It just doesn't sound like a good idea to run into the same problems with the mister core. That's why i personally only test unmerged bin/cues and if that works, i'm fine with it. Anyone can help and write a better heuristic for merged bins, but i will skip on that as it only sounds like trouble for basically no real win

hidden prism
#

How many issues do they get filed about when the logic doesn’t work?

#

I think Robert is (very correctly) thinking about support load

visual depot
visual depot
#

I've used this myself for over 10 years, so I'm curious what other people are doing. My personal belief is that people are sharing the bin files without the cue.... and/or generating new .cue sheets. Obviously you can't just read a bin file and see the pregaps. It needs the cue sheet made during the conversion process.

#

And I can prove why I'm hesitant to think this is an issue with the bin file

#

Here's why

#

I took a bin file from a bin/cue I converted USING CDMAGE from a redump-verified rip. I then CRC hashed it.

I then took all 23 bins and manually copied them bit for bit after each other... no conversion

#

It's the exact same... AND matches the Redump total CRC32 on Redump.org.

The issues aren't the bin files. The issue is people are using wrong/corrupted .cue sheets, is my best guess.

Will fixing this in MiSTer main add another variable for the team: yes. Is that variable controllable? Yes. Tell people to use the right tools: CDMage (and not change settings, as that can be an issue) and it's as simple as opening the cue sheet in CDMage, then clicking save, choose a name and clicking ok. Done. Bin matches redump, and the cue files is properly formatted.

I suspect this website (https://www.duckstation.org/cue-maker/) is the main cause of bad cue files. It's been around for a long time, and is spread all over all kinds of Discords, without explaining that you're supposed to use them with the original bin's. It really should have clarified that.

#

That's why

simple otter
visual depot
#

Oh.... I had mine labeled wrong. lol. I've had it dumped for so long (I edited my original post to put my original thoughts back)

#

it won't

#

The cue sheet already has pregaps

#

It should if you use CDMage

#

This is a cue from CDMage

#

those are the 2 second pregap offsets

#

index 00 for the pregap

#

index 01 for the track

#

The indexes are absolute times relative to the total time on the CD

#

The implementation of merged single bins on MiSTer main is broken

#

no amount of changing the game data will fix the issue

#

someone needs to fix it on the MiSTer end for MiSTer.

#

I was trying to state that if people use the right software, and have redump-verified dumps, the results should always 100% be the same

#

My guess is that 20 years of PS1 dumping has just created a lot of bad dumps, and lost/improper cue files with them

#

15 years ago, it wasn't common for a lot of people entering the scene to know they needed both files. They just think "Rom always one file...derp"

#

And probably trash the cue file. I've done it

#

Then years later they find that fancy cue file generator website on google:

#

It just outputs a single-track cue file

#

it can't tell where the actual tracks are

#

It only makes a 100% accurate cue sheet if the original game only had one track

#

Fun fact... it's intended to be used with multiple bin files

#

yeah THAT is probably a large part of the issue

#

And also if you drag/drop files from windows, it may not drop them in the correct order

#

THIS WEBSITE is most likely the major cause of all the cue file woes imo

#

And it IS popular

#

very popular

#

Spread all over emulator discords

ivory verge
#

I never had problem with this site

#

But it's me

visual depot
#

Well, the issue won't affect duckstation. As Robert stated in his quote above, duckstation will manually add pregaps at runtime if it doesn't see them in the cue sheet

#

and a lot of other emulators do this I imagine

#

They don't actually read them..... because a lot of people have them formatted wrong I expect

#

Duckstation's website I linked will produce the correct cue if the bins are in the correct order, but only for multi-bin setups.

#

But the problem is, there's no verification being done. the webpage doesn't know your files are in the right order. It doesn't know if you have all the tracks.

#

THAT website (and any other tool that doesn't generate them properly or expect the user to know what they're doing) is making the extra variables of user error become a problem

#

Not the actual bin/cue itself

#

If we'd continue to fix the multi-track single bin in MiSTer main, I think it shouldn't take too much work of the support team here, when people have problems, to have them take the separate redump-verified multi-bins, and convert with CDMage

#

They can use redumpVerifier python script to verify the bins, then just open the cue file in CDMage, click save, use default settings and click ok, and name the file... done.

#

Duckstation cue:

  TRACK 01 MODE2/2352
    INDEX 01 00:00:00
FILE "Twisted Metal 4 (USA) (Track 02).bin" BINARY
  TRACK 02 AUDIO
    INDEX 00 00:00:00
    INDEX 01 00:02:00
FILE "Twisted Metal 4 (USA) (Track 03).bin" BINARY
  TRACK 03 AUDIO
    INDEX 00 00:00:00
    INDEX 01 00:02:00```
Original cue:
```FILE "Twisted Metal 4 (USA) (Track 01).bin" BINARY
  TRACK 01 MODE2/2352
    INDEX 01 00:00:00
FILE "Twisted Metal 4 (USA) (Track 02).bin" BINARY
  TRACK 02 AUDIO
    INDEX 00 00:00:00
    INDEX 01 00:02:00
FILE "Twisted Metal 4 (USA) (Track 03).bin" BINARY
  TRACK 03 AUDIO
    INDEX 00 00:00:00
    INDEX 01 00:02:00```
#

Yeah that's my point

#

It creates a proper cue for multi-bins, if the order is correct and it has all the files (it doesn't verify that)

#

But watch what happens if I put a merged bin in there

# 
  TRACK 01 MODE2/2352
    INDEX 01 00:00:00
#

That's it

#

it doesn't match because imgburn is most likely cutting out the pregaps

#

Which would change the hash and it would also product different audio

#

That changing most likely means that whoever was working on it in MiSTer main most likely ended the work with the pregap issue, not sure how to continue since the cue formatting seems to be non-standard...

#

But it's not non-standard... the issue is a lot of people aren't FOLLOWING the standard

#

AFAIK CDMage has been the recommended tool for it well into the PSP days

#

so mid-00's

#

detecting gaps is tough business

#

It is very tough to do it and capture them correctly

#

I deal with this doing secure music CD rips, which i've been archiving for like 10 years lol

#

you basically need the right software with a good CD driver

#

A lot of software do not have this

#

(yes CD ripping software have a driver it installs)

#

nope. Imgburn actually does very inaccurate ripping

#

It just usually doesn't matter for most things, especially audio-based discs like PS1 and music CD's

#

For music, there are a few recommendations, but the most used is Exact Audio Copy, and it needs a very specific set of instructions to dump it perfectly

#

RIP What

#

I will go out on a limb and say imgburn is like the least recommended to do accurate rips if you want it to turn out like redump

#

lol

#

Redump requires you to use their software afaik

#

yeah MPF is probably fine

#

But the bin has to match

#

you can't just use that cue with another bin

#

Basically, to explain WHY it's an issue is simply this: error correction. CD drives have a little feature called error correction that can tell the drive to "correct" a detected wrongfully dumped bit off the disc.

And essentially, for a majority of drives it actually fails to do this right, and sets the wrong bit.

For audio, this actually generally doesn't make the file unplayable, as the only part that would corrupt the file is the header, which is very small. Any part of the audio might even be imperceivable as it could only affect a frequency for a fraction of a millisecond.

The correct way to accurately dump is to disable this feature, and to dump while verifying the data against the original disc, and pulling the data again if it doesn't match, until it's 100% accurate. No software like imgburn even supports doing this kind of dumping.

#

the time would not be affected by wrongly-ripped bits

#

but a bad bit changes the hash

#

and they are very common

#

The result is the same, it's still a 2 second gap

#

What matters is the index 01 points to the same data

oblique frost
#

as an aside, isn't this kind of the point of DIC (what MPF uses underneath for dumps) and using ancient and dying plextor drives for redumping, avoiding the error correction errors?

visual depot
#

Well, it could also be that it's one of the drives that doesn't have bad error correction

#

hmmm.... you're right... weird. not entirely sure

#

but redump doesn't actually store the cue's in merged format

oblique frost
#

yeah I don't doubt it still works. I've been throwing a lot of yennies into the wind lately trying to get a working, approved plextor, so I can actually submit and it bothers me that most of the time is probably won't matter

visual depot
#

where can I download one

#

Yeah, exactly

#

redump verifies against the bins, then uploads the hashes, but it only uploads the split bin cue

simple otter
#

Redump requires dumpers to submit log zips as part of their verification process

simple otter
visual depot
#

I'd expect that considering the bin is matching the verified hash, and the cue sheet is generated from CDMage reading the separate bins, I'm more inclined to believe this is just something MPF is not designed to be used for

#

I've never seen a "PREGAP" option in any PS1 cue file

#

ever

#

And I've seen a lot of PS1 cue files on redump sets

#

Oh. lol then yeah don't trust imgburn

#

I've said that

#

It can't detect pregaps properly

#

I'm going to bed

simple otter
#

side note: MPF's .img cuesheet is identical to the one generated by converting an MPF multi-bin+cue to a single bin+cue with CDMage

#

It is safe to say that the .img file and _img.cue(found inside _logs.zip) from MPF are equivalent to .bin and .cue files resulting from merging multi-bins with CDMage

Unfortunately does not much improve the fact that MiSTer does not handle multi-track single-file bins without converting them to CHD first

visual depot
#

CDMage released in 2001. Lol

simple otter
#

I am only comparing insomuch as there is no point in me joining bin files and creating cues when the bins and cues are already joined and generated

visual depot
#

If the bin matches the redump hash and the cue file looks like the one from CDMage they are 100% the same, bit for bit.

#

When dumping yes. But you can't use mpf to verify dumps you already have can you?

#

Yeah...

#

Mpf is only designed to dump and verify at the same time.

#

Let's face it: we can't expect to lord over everyone using MiSTer to pull out their discs and dump using the very concluded dumping instructions of MPF. Not everyone has the attention span.

#

Verifying with the redump verifier script and the multibins, and then using CDMage to convert is both easy and 100% same result every time, and accurate to redump's "total hash".

#

Not sure why the python script doesn't include totals in the checks. Maybe it's not in the dat

#

I looked at the mpf dumping guide page and immediately feel overwhelmed

#

Doesn't matter how little steps there are. People will look at that and avoid it

#

Anyway bed. Night

tropic stratus
#

this was like 4 screens of technical discussion to do it right and honestly i couldn't follow everything...How do you suppose the typical user having dumps from either 5 different 20 year old tools or 5 different shady websites be able to follow the rules?

#

In any case: feel free to improve the singlebin code in mister main. The current one is more broken than it should be so it can at least follow the general case. But sorry, i will not debug any CD DA issue due to merged bins. It's not worth the time if the multi bins work

tropic stratus
#

i would guess that 99% here in the channel rely on redump verified multibin/cue or chd. This is the first big discussion about this topic in the last 1.5 years

#

but i don't see the problem: it's open source and i'm very happy if you improve the merged bin/cue HPS code

#

the problem is that the original author of the merged bin code is no longer there

#

the multi bin code is from sorgelig, the chd from Zakk

#

i don't complain that you report the issue. It's known that merged bin HPS code is not good. I just tried to explain why it is the way it is and why there was no big interest in improving it

#

There are also a ton of other formats that are usually supported by many emulators, but it was decided against it....to some degree this also happened with merged bins

#

you wouldn't even find it inside the core, as it's in mister main 😅

dusky martenBOT
# 
static int load_cue(const char* filename, toc_t *table)
full nimbus
tropic stratus
#

yes it's c

hasty jay
#

something slightly related: I have a patch for mame's chdman tool to convert chd to multi bin instead of single bin. I wonder what the chance is to get that merged (I'm not sure what the benefit would be for anyone but us)

oblique frost
tropic stratus
#

also 280-314

simple otter
#

I deal with perl mostly so this isn't too far off from what I'm used to, it's fairly legible

oblique frost
#

sympathetic for your dealing with of perl :p

tropic stratus
#

i guess that's what we all share here. While some(many?) will inevitable have a technical day job, we are mostly here for hobby 🙂

wanton crown
#

so a de 10 nano bought one month ago apparently did not show the behavior I was having with Clock Tower - Ghost Head.
Tomorrow I'll be able to test on a third de 10 nano board (one that I have at work and is quite recent), if it works it'll mean there was a "bad" batch of de 10 nano around last october/november

#

and I'll just buy a new de 10 nano board once again ....

quick gyro
#

are you building a MiSTer House? 😅

wanton crown
#

well, I tend to sometimes be at my parents and sometimes at my place, and I have two CRT setups with one mister each

#

and finally, one at work so we can play some multiplayers games, especially on the PSX core during breaks

#

if I can prove it's the de 10 nano board (which by now I'm 99% sure it is) somehow, I'll just get rid of it and give it for free to a workmate or something, or better, someone who really needs one

quick gyro
iron abyss
# wanton crown if I can prove it's the de 10 nano board (which by now I'm 99% sure it is) someh...

Do you have a pic of the issue, just curious to see what it looks like. The DE10s definitely aren’t flawless, I had one of the GPIO pins get stuck on one of mine despite being in a case and no ESD. I simply used the button wired to that pin once after months of it working flawlessly and then it was forever stuck low. Having two fail seems pretty unusual though, we’re you able to try swapping the ram? You could do diode check on both GPIO headers with a multimeter (skipping power/gnd) and see if some pins are sad/borderline like what happened to mine.

lilac scaffold
#

@iron abyss #1046940919607345272 message

wanton crown
#

@iron abyss it all started here
#1046940919607345272 message

#

Tl;dr artifacts in a single game, all others work flawlessly and never had a single issue with the psx core. Started with the 2022.10.21 update (everything before that doesnt exhibit the issue), but that is only for me on TWO different mister fpga setup, all bought 1 month apart last october up to november.

I'm not going to annoy this channel anymore with this matter, but at this point there has to be a very weird edge case or something with one of the de 10 nano board revision around that time.

iron abyss
#

That’s wild, just scrolled through it

#

Did you try removing the IO and just doing bare DE10 + ram with hdmi out

wanton crown
#

What annoys me is that if we start getting "bad" batches, it makes me lose a bit of faith in the whole mister project. I had encountered such issue with the cps 2 core a while ago, nobody cared or even tried to reproduce, so as I was the only one the core developer didnt bother as well. Can't blame any developer at all, we're just leechers anyway. If terasic is doing such revisions without maintaining a proper revision log (can't see any recent one on their website), it just seems to become a chip lottery. If that's what it is we're all at lost, developers aren't going to invest every single new impossible to reproduce report because of a board revision they can't even easily have access too

wanton crown
summer sage
#

it's actually pretty unlikely it's the hardware at all

#

more likely to be some setting you don't realize

random gust
#

Let's get photos of the settings then.

summer sage
#

I'd be frankly shocked if you had two pieces of hardware defective in the same way twice in a row.

wanton crown
#

Photos were posted, settings shared, sd cards changed, fresh mister setup through update script, every single time the issue was there.

summer sage
#

stuff to check for:

  • Make sure you're using fast enough data delivery, whether it be network or sdcard
  • Make sure you aren't running any third party things on the HPS that might interfere
  • Make sure you aren't overclocking HPS or doing anything else that might compromise it
  • Make sure your disc image is valid, matches everyone elses sha1's, etc
  • Check your settings and core version against everyone elses
wanton crown
#

Tbh I don't think it's worth polluting this channel more with this matter, I and people here already wasted a lot of time trying a lot of things while it seems obvious it's related to the de 10 nano board

#

Disc images were verified, dip switches on the de 10 were verified, sd ram was verified (everytime on both setups), made the de 10 nano board naked with the bare minimum for the core to work, took 3 different sd cards and made a fresh install everytime, power supply was checked...

summer sage
#

Okay well then don't go dropping "I'm going to lose faith in the mister hardware because im getting 6 bad de10's in a row" when it's almost definitely not that.

wanton crown
#

Well it's simply my opinion based on a very rational reasoning

summer sage
#

getting deterministic results from one defective fpga are rare, two in a row would be astronomically unlikely

wanton crown
#

If a bad batch happened for me over the span of a month, it surely can happen to anyone. Not everyone has the chance to be able a ton of de 10 nano

summer sage
#

they test fpgas as well before they determine their grade

#

these are the highest grade of them, rated for higher temperatures

wanton crown
#

I agree with you, hence why I had "" to "bad" batch, to me it could boil down to ddr3 revisions, all the board I see have different ddr3 chips

#

I'm not saying both my de 10 nano operates bad, I'm just saying they obviously differ from all the testers so far.

summer sage
#

it's almost certainly some external factor

#

you just have to find what it is

wanton crown
#

It will be ruled out tomorrow night once I bring my most recent de 10 nano board from work, if it shows the same behavior then I would tend to go back to the "exteral factor" theory

summer sage
#

vauge accusations of "bad batches" without definitely determining what your problem actually is just starts annoying rumors that linger forever

wanton crown
#

I will employ "batches that cause unwanted behavior on the psx core for a single damn game" then

summer sage
#

it's much more likely, by a large margin, that it's something local to your setup than anything to do with the de10

wanton crown
#

I'd be willing to bet several thousands of $ and have someone check for us that indeed my boards differ while running the core with a bare de 10 nano board and all the sd ram sticks they want

visual depot
#

This is the only part that I think matters to read

#

TL;DR the hash matches the "total" hash on redump after merging with CDMage OR just bit for bit joining all files together manually

abstract drift
#

for a long time xstation recommended merged bins for performance reasons

#

and psio still needs them

visual depot
#

my point is it's the exact same data. The issue people have with the files, I expect, is improper cue files

abstract drift
#

yeah

visual depot
#

So does CDMage. Looks like this tool may create the correct bin/cue too

#

My question is if the bin turns out with the correct hash though

abstract drift
#

the bin should be really hard to get wrong afaik

#

its just the tracks written one after the other to a big file

#

that binmerge tool has a split mode so you could merge/split and recheck hashes

visual depot
#

Not asking you to add the code @tropic stratus. Haha. Never meant to imply that, but it really shouldn't be adding any extra work on your front. If people report issues, then having them collect a verified redump dump as normal and converting with CDMage or binmerge above should net them a perfect merged bin that is identical to the hash on redump

summer sage
abstract drift
#

its the people trying to convert like, psp images i think they are? to run on psx that always gets me

scenic reef
summer sage
#

"fix" is the wrong word

scenic reef
#

sure

#

improve

#

it's something like only a couple percent more cpu usage, not even something that would mess with a very slow ARM processor, slower than the de10-nano's

summer sage
#

the gains are there, but the difference between no compression and old compression, vs old compression and new compression is large

scenic reef
#

sure

summer sage
#

the new stuff wouldn't really be that big a deal, relatively speaking

#

not worth breaking it

scenic reef
#

it would make a significant difference... if storage was still expensive per TB 😛

#

but it's not so it kinda depends on the consumer of it

summer sage
#

even if it was a 25% reduction it's still not a big deal

#

and it's probably not that much

#

basic old compression is around a 50% reduction

#

chances are you'd get maybe 60% instead of 50%

#

is that worth having tons of devices not work anymore? not to me at least

#

much better to focus on widespread adoption which means stable reliable format

#

but you know what, it's their format, but personally, I hope they dont change the compression

scenic reef
#

currently libchdr uses huffman and it's not using any standard one like zlib or lzma, etc... right?

summer sage
#

the de-facto compression method for windows is still zip

scenic reef
#

well huffman is standard

#

but i mean like some open source driver/lib

summer sage
#

if that tells you anything

scenic reef
#

yeah zip is lame lol

summer sage
#

but it's good enough

#

and everything supports it

simple otter
wanton crown
#

Mister main is updated to latest, like the psx core. I would assume everyone who try to reproduce the issue went with a similar setup.

summer sage
#

I've seen problems like this caused by bad power switches

scenic reef
#

if main can be improved to handle the multiple track single bin non-redump ones, it would probably have to have a bunch of complicated logic to essentially make up for the shortcomings of bad conversion tools

#

so it's not much of an improvement

simple otter
scenic reef
simple otter
scenic reef
#

ah yeah

#

they may be supported one day if the combined bin was done correctly in a way that the core might accidentally support, but i assume it's on purpose to not support it since there are like 2 decades of rips that have been merged by different tools and different versions of the same tools floating out there, so it causes false positives in issues

#

emulators supporting bad rom dumps and mangled roms is bad practice

wanton crown
visual depot
abstract drift
#

i tested merged bins when it first went in and it does at least work somewhat

visual depot
#

games boot

#

but when it tries to access the other tracks, it will generally fail with loops

abstract drift
#

ridge racer and wipeout play cdda

visual depot
#

which is generally just the music

abstract drift
#

or they used to

#

thats what i was testing with

visual depot
#

From what Robert said, it's due to the pregaps not being read properly. They are supposed to be defined in the cue, but a lot of bad software or bad cues don't have them defined, and thus the dilemma of how it should be handled on MiSTer was born

abstract drift
#

my cues may be correct as it turns out

#

they work on an actual psx afterall

visual depot
#

They are just bit for bit joined from the separated bins

#

So they have the same redump verification process

#

as long as the dump is redump verified, there should be no qualms when using CDMage or Binmerge

#

period

#

The hash for the merged bin is even on redump.org's database

#

It is calculated during upload alongside the separated files

abstract drift
#

makes sense, when they're done correctly they work, when they're not they don't

summer sage
#

they aren't fundamentally different

#

just slightly harder to debug

#

I dont know if any official groups release verified merged sets

visual depot
#

I've seen them before

#

I understand the concerns are extra workload being placed on Robert and the team. But I don't see it being anything past what is already asked of users: using a verified redump bin (you can calculate the crc32 of the merged bin and compare to the "total" hash on redump.org)

#

If it isn't, then the user should get a proper verified dump and merge with binmerge or CDMage

#

that's all there really is

#

I mean, you guys normally ask if the dump is verified anyway, so that part doesn't change

summer sage
#

no, it's not that

#

it should really definitively work or not, and honestly it's no so different that there's an excuse for why it shouldn't work

visual depot
#

I guess the only thing is the cue file, which a very easy look at would show if it's bad

summer sage
#

I just don't think it's especially a desireable format

#

of the three typically dealt with, it's the least so

visual depot
#

I think the only reason why it would be desirable is that a lot of people store them in that format for PSIO and other tools and emulators that require it

summer sage
#

usually I hate splitting files up, but cd images are historically quite error prone

visual depot
#

merged bins, thankfully, are 100% the same data as whatever is put into them

summer sage
#

yes, I get it

abstract drift
#

xstation recommended merged bins because they found seeking inbetween different files was slower than inside the same file

summer sage
#

they are just catted

abstract drift
#

i kinda don't expect mister has the same concern though

visual depot
#

but no I totally understand your concerns Kitrinx

summer sage
#

I feel like chd is the future

visual depot
#

I agree. CHD seems like a good format to base off of... the thing is CHD isn't able to reverse back into split bins I believe

#

It will spit out to a merged bin if you un-chd the file

summer sage
#

why not?

#

just because nobody has made the tool? there's isn't any reason it shouldn't be able to

abstract drift
#

you can unmerge that bin at that point with binmerge

visual depot
#

fair yeah

#

basically CHD is merging it inside to my knowledge

summer sage
#

it's a shame it doesn't support subtrack stuff, but im not sure how much that really matters

visual depot
#

It's just lining the data up and using lzma (I think) compression

summer sage
#

I guess the 12 or so CD+G fans are sad

abstract drift
#

chd is storing the whole datastructure in a completely different way

visual depot
#

I guess the CDDA gets FLAC'd

#

yeah

abstract drift
#

its got like ... stuff in it to aid random seeking

summer sage
#

yes, it sets them up for more efficient data access

abstract drift
#

like i think it might be a b-tree or something

summer sage
#

what things require subchannel data?

#

i've heard about it for years and years and never really figured out why we need it

abstract drift
#

mainly that copy protection thing isn't it?

summer sage
#

I genuinely don't know

abstract drift
#

libcrypt?

visual depot
summer sage
#

which games don't work then?

abstract drift
#

spyros the classic

summer sage
#

why does it work on mister?

abstract drift
#

always EU versions, no libcrypt on USA

#

we have support for .sub files

visual depot
summer sage
#

ah so it just side loads the subchannel data

abstract drift
#

a pack of all of them is only like 100kb or something so i think its bios getter just gets them all

visual depot
#

Apparently mainly affected PAL and NTSC-J

summer sage
#

I guess it's not that huge a concern then, but it would be nice to have them stored in there somewhere

abstract drift
#

xstation do the same trick - all the .sub files are baked into the firmware

#

yeah would be nice if chd supported them

visual depot
#

FF9 seems to be a popular example for the Japanese region

hasty jay
abstract drift
#

have you tried an MR for that? sounds generally good

#

idk how fussy the chd team is though, part of mame right

hasty jay
#

I have a branch, I can make an MR, but don't really know how to justify the change

abstract drift
#

makes the chd process 100% reversible

hasty jay
#

what if you make a CHD file from a merged bin 😉

abstract drift
#

then don't use the split option 😛

hasty jay
#

then the reverse would be broken

#

well, with my patch it's not optional, but we could extend it to optionally split or merge

#

compiling mame right now (which takes forever) to see if it still works

#

why can't they just let me do "make chdman" :/

visual depot
#

Honestly I don't think making a chd from a merged bin is a good idea anyway. Does it pick up the tracks properly from that?

abstract drift
#

it'd just read the cue anyway wouldn't it?

visual depot
#

I mean it's totally possible for the resulting CHD to be 100% identical if it reads the CUE

#

I'm just not sure if it does

abstract drift
#

it has to to find out what the tracks are

#

like, audio/data/etc

visual depot
#

the cue defines that though

#

And I'm pretty sure there's always only one data track

hasty jay
visual depot
#

honestly I find Duckstation to not handle it well

abstract drift
#

only one data track is only a feature of psx discs in specific afaik

#

cd's can do all sorts of weird stuff

visual depot
#

If people use correct rips, cue files should be a non-issue, and Duckstation allowing you to generate a cue file on their "generator" site without even verifying the rips (order, file count, etc) is a large part of this issue honestly

#

I really think making that website without a disclaimer to be sure you have good rips (and especially not to use merged bins) was a horrible oversight

#

it doesn't even tell people that putting a merged bin on that site will always generate a wrong cue file

#

like, seriously. That needs to be mentioned

abstract drift
#

is it actually by the duckstation dev?

#

you could open a (polite!) github issue

hasty jay
#

During the implementation of libcrypt support for the PSX core I found out that all this subchannel data in the end just contains 16 bits of information

dark mountain
summer sage
#

it seems like everything about them is stalled

hasty jay
#

The game can just see that the subchannel data is valid or invalid and there are fixed positions where it checks. In fact we only transfer these 16 bits to the core.

summer sage
#

but, that's kind of okay

#

maybe some other systems care more about subchannel data?

hasty jay
#

yes, possible

summer sage
#

it doesnt seem that it matters to sega cd or tg16 cd

abstract drift
#

psx is the only one i've heard of caring

summer sage
#

not sure of saturn

abstract drift
#

oh maybe xbox360 actually

summer sage
#

I guess it probably is not very significant and this is why it was an oversight in the format

hasty jay
#

I guess the reason it's not stored is that it's completely deterministic what you should get back. Unless you have a scratch on your disk or someone messes up the data on purpose as a copy protection 🙂

summer sage
#

in theory

#

couldn't it just be added as an extra cd track that software was aware of?

abstract drift
#

yeah i think so...

summer sage
#

since chd can contain any type of file, it seems like it can just get tacked on

abstract drift
#

not sure if games might puke on the extra track existing though

#

like if they ever seek to last track

summer sage
#

it's a matter of how you deliver data

#

if it's flagged correctly the data delivery can do it properly

#

with chd there will always be an interpretive layer between the data and the thing that needs the data

#

I dont know the format well enough to know how it stores discs vs files

#

but can't it also be used for like, mame roms?

#

maybe you can literally just mix a file in there with the disc

abstract drift
#

it can store hard disks and stuff i think

#

so pretty flexible

summer sage
#

then mister's thing can yoink the file out and deliver it

#

@manic citrus do you know?

abstract drift
#

doesn't that end up creating chds that only work on mister though?

summer sage
#

im not sure what something else would do, but possibly just ignore the file it's not looking for?

manic citrus
#

is this some subchannel data stuff?

#

(sorry, that's a lot of scrollback)

#

if so, chd already can store that

#

it's how CD+G works for megacd

abstract drift
#

sounds pretty extensible

mystic violet
#

god you guys chat a lot

wanton crown
#

passion and obsession is the way to improvement

manic citrus
#

note: most of the common psx dumps are unlikely to contain subchannel data

abstract drift
#

apparently chds support inheritance too, for multidisc games. not sure if thats hooked up on msiter though

manic citrus
#

you have to explicitly pass in the parent file name

#

there's no sane way to do that

abstract drift
#

couldn't you just assume it'll be like, foldername.chd ?

manic citrus
#

which one is the parent?

#

also it's not for multi disc, it's for discs with shared data. which may or may not be multidisc

#

more likely it's for different regions that share 99% of the same data

#

also no dumps use that feature

abstract drift
#

oh i didn't think of regions

#

i was just thinking of multidisc games

#

also you'd only be saving like, a few GBs really

manic citrus
#

you could potentially use it for patches, too

abstract drift
#

you could design a convention for it i guess but it might not fit in well with peoples already existing folder structures

manic citrus
#

but yeah, subchannel stuff is doable. but I think the psx scene just needed it for that weird EU copy protection and they did it via those separate files which mister already handles

#

I don't even think redump has them? although to be fair I skipped the entire EU set

abstract drift
#

redump is where they come from

#

how does it work on megacd?

#

like how do you actually embed the data into the chd? maybe it already works on psx core and no-one ever tried it 🙂

manic citrus
#

megacd has a way to send that data

#

in a chd they are just globbed onto the end of each sector

visual depot
#

So even tips without them work since it just reads them from the firmware

abstract drift
#

i thought xstation did that but psio actually doesn't

visual depot
#

Sorry xstation

#

Lol

abstract drift
#

you have to get patches for psio

visual depot
#

My point is, they are tiny and not really copyright controlled I'd say

abstract drift
visual depot
#

But doing that on mister would be weird.... Not sure that could be done without saving a db on the SD card which seems a bit.... Weird

abstract drift
manic citrus
#

it's already done on the mister

visual depot
#

Wait it is?

abstract drift
#

yeah embedding them in the chds would just be neat

#

yeah bios getter downloads them all i think

visual depot
#

Oh lol I misunderstood that part in the conversation

manic citrus
#

I have no idea if the redump converted to chds on archive even have the sbi data re-embedded into them

summer sage
#

well, there you have it

#

"I was the turkey all along"

manic citrus
#

as god as my witness, I thought turkeys could fly?

summer sage
#

anything can fly if you throw it hard enough

mystic violet
#

"I believe I can fly!!"

full nimbus
#

Shoot a turkey with a big enough projectile and it flies

summer sage
#

wild turkeys can certainly fly

manic citrus
#

Yeah, because chdman handles single bin correctly

summer sage
#

@manic citrus can someone make a chd with the subchannel or does our main code not know how to deal with it?

visual depot
#

You can convert to chd from multibin directly lol

summer sage
#

im pretty sure once it's a chd the original format of the data is meaningless

visual depot
#

Yeah pretty much. It translates the data losslessly either way. It's the same data whether it's multi or merged.

#

How are we not.

#

I've converted multibin to chd directly.

#

Have you tried it.

#

Lol

dark mountain
visual depot
#

You literally just point chdman to the cue file regardless.

full nimbus
#

I think after 2 days this should be taken to #offtopic its only tangentially relevent to PSX and its drowning the channel

summer sage
visual depot
#

Yeah honestly I think it's pointless to keep bringing it up at this point. kitrinx already said she feels it'd be good to finish the feature.

dark mountain
#

Okay, probably a fix since when I'm retired of the scene, I've seen some move too on the Saturn and DC sides.

visual depot
#

Just have to wait for someone to fix it. In the meantime use multibin or chd

#

They meant us.

#

Lol

summer sage
#

im not really sure why you're even still talking about it

#

in summery
-yes combined bins are broken
-yes it'll eventually get fixed
-chd doesn't give a flying fuck where it's data came from and will spit out whatever you want if converting back

visual depot
#

I think there's no reason to discuss it anymore lol

full nimbus
#

And no reason to do it in PSX. Create a New thread about CHD and leave it there

simple otter
full nimbus
#

then still no reason for it to be in PSX

#

Create a thread and argue away in there

dark mountain
#

I've done the whole set back then, most probably 3-4 years ago and never had a problem with it nor other users who taken those files from other sources.

summer sage
#

im more interested in getting the subchannel data into chd's for the handful of copy protected things

dark mountain
#

Same.

summer sage
#

not that it seems like much of a concern for mister

#

it would still be nice, especially since it's already possible

dark mountain
#

The most knowledgeable person on those subjects could be claunia, the creator of Aaru which is the heart of the whole Redump/MPF system since years.
Or RibShark, maybe.
But they should probably have better things to do right now, like supporting exotic filesystems and integrating laserdisc dumps in the (near ?) future.

visual depot
#

Would adding that to chd be possible without requiring all software to update to support it?

#

Like would it break current chd's from working.

dark mountain
#

That was part of the discussion.

#

Maybe something like with iNES headers, the 2.0 revision ROMs are compatible with older emulators.

visual depot
#

I gotcha.

#

Headers sounded like something that would work but I asked because idk if the chd format can add a header like that.

#

Obviously implementing an addition that doesn't break references to other things would be ideal.

dark mountain
#

One of the first ideas when creating/updating file formats for preservation, don't break back compatibility.
But sometimes you will have and probably add also technological debt.

visual depot
#

Yeah... It depends on if the structure itself was designed to be updated.

#

I understand

#

Just wasn't sure if chd was that way, was all

tropic stratus
summer sage
#

oh I see

tropic stratus
#

or was it for other CD cores?

summer sage
#

the others already have what they need, unsure of saturn

#

probably a waste of time to think about, not many people play pal games anyway

#

and none of the existing rips include this

tropic stratus
#

as markun said: the "subchannel" is currently downloaded from mister main to the psx core in 16 bit together with the CD "header" information 🙂

#

that's all that is really needed. So far no game has been found to have a problem with it. I played several of them

#

i found this quite important and i'm very grateful markun found this nice and tiny solution. Before i planned to download way more .sbi data to the core

ivory verge
#

Yes , I only play PAL versions of games and I haven't had a problem with any of them

wanton crown
#

same

hasty jay
#

@wanton crown I was curious what your nick means until I converted it to decimal 😄

ivory verge
#

elite 🙂

visual depot
#

Only select PAL and Japanese games seem to be reported to use it for copy protection

#

Final Fantasy 9 Japanese version seems to be one of the major noticeable ones

dark mountain
#

FF IX Japan isn't protected.

visual depot
#

I'm going by statements here

#

Sub channel reading basicly allows you to get around many protections without getting a PPF file to patch the emulator. This is mostly interresting for NTSC J (japan) and PAL (europe, asia) users, as these versions of the game often come with such a copy protection (the most popular example would be Final Fantasy 9 here).

dark mountain
#

Only the PAL versions of Final Fantasy IX are protected.

visual depot
#

Ah... alright

#

Lack of clarification on the post lol

#

Sorry

dark mountain
#

And no game in Japan have the LibCrypt protection, that's an error in the documentation.

#

In Europe, it was common to see people very rapidly with modchips and games on CD-R, there was even ads on major video game magazines for modding systems just after a PS1 game review.

#

Officially, for removing region lock.

ivory verge
#

Yeah, I remember it well 🤣

#

I also remember a review of a game that never came out 🙂

dark mountain
#

Everyone in my middle school had their burned PS1 games binder into the schoolbag during recess.
And it was time for exchanges, so cracktro'ed games were easy to find too.

ivory verge
#

I had a nice Japanese edition of Parasite Eve II , the discs were pressed

#

I am not a supporter of piracy, but that's how it was at the time

dark mountain
#

Just a reminder of those times.

quick gyro
#

Also Bleem was in stores super fast. I remember some friends who had a pc telling me that the psx will fail because the games can just be played on Bleem.

bright yacht
#

Trying to see what's on my memory card... using snax and a multitap... Bios does not detect it. But ingame it works. Any clue?

still palm
#

try an unstable build after Jan 9th

#

or maybe the latest release build

bright yacht
#

Ok. I will try tomorrow and report back. Thanks.

ivory verge
#

Yes , mem cards work well with the latest release , your fix is there, thanks @still palm 🙂

wanton crown
#

Just reproduced the bug on a third mister fpga...

#

Im at loss

random gust
#

Wasn't there someone who recently got a DE10 Nano? @wooden rose was it? Might just have to see the results from tests on that.

wanton crown
#

He couldnt reproducr

#

This is really insane

#

Thats yet a third buid with another 128 mb sd ram etc

ivory verge
#

So I don't think it's de10 nano's fault 🤷🏻‍♂️

forest trench
#

When trying a stock install from scratch, are you using a 100% untouched .ini?
Or are you changing display options for your setup?

timber veldt
#

Well, reasons

iron abyss
timber veldt
#

You have to replay the game

iron abyss
#

Yeah was going to suggest that, play through to that point again, or restore someone else’s normal save game near there and then go see if you can get the issue

wanton crown
#

Yes same save state shared here

#

Replayed and it occurred as well

ivory verge
#

Ok so to summarise, on three de10 nano you have the same error and no one else can reproduce it....

tacit fox
#

Not the de10 nano for sure

#

Did you try with stock power supply?

wanton crown
#

Yea

#

Buy they were all bought between september and november

#

All setups ate completely separate, i.e difgerent ram, power supply etc

iron abyss
#

How much of a sample size has there been for reproducing it? Just a couple other people? Maybe it’s like CPS where it’s tough to get into the bad state

#

Though I guess it’s everytime for 0x539

ivory verge
#

Many people have tried to reproduce this, it only occur for 0x539 at the moment

#

If I needed a third de10 nano I would gladly buy one back from 0x539 🙂

iron abyss
#

I’ll trade mine with GPIO1 pin 13 stuck low 😉

ivory verge
#

Repairable😉

iron abyss
#

“For parts or repair” or “untested” I guess is the better modern listing for “this is absolutely broken”

ivory verge
#

Haha sure 😎

wanton crown
#

Im buying a 4th de 10 nano

#

Fuck ir

#

Fuck it0

#

Fuck it *

ivory verge
#

Don't do it...

iron abyss
#

Isn’t there some kind of hw diagnostics tool from Terasic, maybe try that to further rule out hw differences. Really is a mystery if settings are stock and everything’s mentioned has been tried. But given Natrox couldn’t rep it seems less likely the hw theory has any merit

#

Where’d you get your sdram? Maybe try to buy or borrow an older revision or from someone else?

wanton crown
#

Havent found someone who bought one from terasic in september and october

abstract drift
#

this is fun

#

have we tried different displays/hdmi cables?

wooden rose
#

You need to get rid of any factors that could influence the result, aka try an absolutely barebones setup preferably at someone else's home

#

If you can't reproduce it then I'm sorry to say, but your home is cursed by the glitchy games ghost

shut escarp
#

dibs on one of the 'broken' ones

wanton crown
#

This is at work

#

So yet again a brand new setup, different power supplies, sd ram etc
They should all be good. I do not reuse a single damn thing between all these setups but the issue is always there

#

And yes each time the de 10 nano board is naked, with just the sd ram plugged in and just the necessary things to boot the psx core

abstract drift
#

maybe your cd image has some bits flipped in a way that is generating a hash match despite being flawed?

#

its incredibly rare - hash collisions, but we've ruled out most likely things

iron abyss
#

It turns out 0x539 issue is more interesting than majority PSX games 😉

oblique frost
#

0x5 lives under constant solar flares

wanton crown
#

I'm so disgusted