PSX Compatibility List

[MostroW]

Tekken 2 (USA) (Rev 1).chd MD5=FAB7C1DFA7C6AB3160A851E27ACFC2B3

This game worked fine for me, intro without hitches or missing sound, ingame worked just as good.
But i've converted my whole library to CHD format so i can't say if it would match up against the redump version without converting your dump to CHD format.

[chunky_tesco]

MD5 (Tekken 2 (USA) (v1.1) (Track 1).bin) = 14c6b7308e1bf796fb36e523adde34d7
MD5 (Tekken 2 (USA) (v1.1) (Track 2).bin) = 9f64359d59fa921a68dea3300bd54ac2
MD5 (Tekken 2 (USA) (v1.1) (Track 3).bin) = 131bb82eb1ba56b4991091f1717770a2

MD5 (Tekken 2 (USA) (v1.1).chd) = d9a459f08c1ff9a3055b0bd933f0e0fd

so the bin's match redump MD5, but the chd doesn't match yours

[chunky_tesco]

i've got fab7c1dfa7c6ab3160a851e27acfc2b3 now - I ran chdman on windows rather than MacOS, but still the same issue. Could be some jumper settings on the DE10?

[MostroW]

at this point i can't say, the only thing left for me to add, is that i run the game of an usb connected SSD.

but if you would be running of sd card of a conventional hdd that shouldn't be a problem as the ps1 had a 2x cd drive which was 300Kb/s at best?
also i would've expected other games to have similar issues, so i don't know if that is the case if you have any other titles like tekken 3?

i did my conversion with cdhman on windows as well, so i can't see that being the issue, also your MD5 checks out, the checksum should've been different if the data would've been different?

if you got a spare sd card capable of containing a basic MiSTer setup and add the latest ps1 core together with the game to test, i'd advise it to see if either the core on your current card or the card self might be at fault?

[chunky_tesco]

Tekken 2 (USA) (Rev 1).chd MD5=FAB7C1DFA7C6AB3160A851E27ACFC2B3

This game worked fine for me, intro without hitches or missing sound, ingame worked just as good.
But i've converted my whole library to CHD format so i can't say if it would match up against the redump version without converting your dump to CHD format.

can you post the MD5 hash for your boot.rom?

[chunky_tesco]

this is strange, as Tekken 3 works fine, as does Tekken 2 (PAL version). It's just the NTSC version that has the problem.

I've changed SD cards (I upgraded to a larger card) and that didn't help

Could it be something like faulty RAM? I'm really scratching my head with this.

at this point i can't say, the only thing left for me to add, is that i run the game of an usb connected SSD.

but if you would be running of sd card of a conventional hdd that shouldn't be a problem as the ps1 had a 2x cd drive which was 300Kb/s at best?
also i would've expected other games to have similar issues, so i don't know if that is the case if you have any other titles like tekken 3?

i did my conversion with cdhman on windows as well, so i can't see that being the issue, also your MD5 checks out, the checksum should've been different if the data would've been different?

if you got a spare sd card capable of containing a basic MiSTer setup and add the latest ps1 core together with the game to test, i'd advise it to see if either the core on your current card or the card self might be at fault?

[MostroW]

faulty ram then i would've thought that more games and more cores would have shown issues beside only tekken 2.

the strange thing that only your US tekken 2 has this and nothing else.
the image md5 checks out, i'm afraid i can't suggest anything more besides this, i've got the same image (on a MD5 checksum) as you, and the game runs fine for me.

[chunky_tesco]

I've found the issue, it was down to a corrupted/bad save that I downloaded and coverted to the memory card.

[caffeinekid]

Do any games still not run at this point? It seems to be fantastically compatible in my tests now.

[Caldor]

Do any games still not run at this point? It seems to be fantastically compatible in my tests now.

I am very surprised that Robert can keep finding things to fix. Even now its pretty much daily that he adds new updates to this core, often fixing very specific things or optimizing something to use maybe just a few CPU cycles less to make it faster, or have something wait for something else.

Some of the recent update changelogs:

- DMA/RAM: remove 128bit data bus, deliver data to DMA with every 32bit word -> lower ressource usage and lower latency
- DMA: due to seperate ram ready signals for DMA, DMA does not have to wait for pending data before switching back to CPU -> lower latency when DMA ends
- DMA: unify blocklength calculation for DMA to/from SDRAM
- Savestates: wait for valid spot before going into reset/savestate load
- Savestates: add retry logic if reset ends in permanten pausing or permanent not pausible situation
- DMA: early stop ram autoread as soon as all required data is requested
- DMA: add separate state for DMA start and reduce initial overhead by 2 cycles
- RAM: ram reqprocessed only used for dma now -> removes risk of CPU reads influencing DMA reads
- Savestates: fix reset when loading .exe files
implement 3 different turbo speeds:
- Low: slightly faster ram access
- Medium: additionally 50% cache hit rate
- High: full cache hit rate + speedups for DMA, GTE and CPU

Just amazing that he can keep doing this.

[FPGAzumSpass]

Do any games still not run at this point? It seems to be fantastically compatible in my tests now.

Yes, some are listed in the github issue tracker.

Most of them have a workaround, so totally not working are only very few.

[AmintaMister]

Do any games still not run at this point? It seems to be fantastically compatible in my tests now.

I am very surprised that Robert can keep finding things to fix. Even now its pretty much daily that he adds new updates to this core, often fixing very specific things or optimizing something to use maybe just a few CPU cycles less to make it faster, or have something wait for something else.

Some of the recent update changelogs:

- DMA/RAM: remove 128bit data bus, deliver data to DMA with every 32bit word -> lower ressource usage and lower latency
- DMA: due to seperate ram ready signals for DMA, DMA does not have to wait for pending data before switching back to CPU -> lower latency when DMA ends
- DMA: unify blocklength calculation for DMA to/from SDRAM
- Savestates: wait for valid spot before going into reset/savestate load
- Savestates: add retry logic if reset ends in permanten pausing or permanent not pausible situation
- DMA: early stop ram autoread as soon as all required data is requested
- DMA: add separate state for DMA start and reduce initial overhead by 2 cycles
- RAM: ram reqprocessed only used for dma now -> removes risk of CPU reads influencing DMA reads
- Savestates: fix reset when loading .exe files
implement 3 different turbo speeds:
- Low: slightly faster ram access
- Medium: additionally 50% cache hit rate
- High: full cache hit rate + speedups for DMA, GTE and CPU

Just amazing that he can keep doing this.

We're talking about a genius...

[TobyRieper]

So the PSX core is more accurate / compatible then the PSIO and possibly Xstation?
Wow. I'm so chuffed I repurchased a Mister. Funnily enough... the first core I played was Tekken 2 (NTSC) Played great.

[FPGAzumSpass]

Not sure about PSIO, but XStation is likely more compatible as it only has to exchange the CD part and does a great job with it.

Hard to compete with rebuilding everything against "only" CD
(CD is maybe the hardest part, so hopefully that doesn't sound rude)

[TobyRieper]

Not sure about PSIO, but XStation is likely more compatible as it only has to exchange the CD part and does a great job with it.

Hard to compete with rebuilding everything against "only" CD
(CD is maybe the hardest part, so hopefully that doesn't sound rude)

No probemo, amazing work with the PS1 core. Hats off to you. Jusy playing through The X-Files (USA) and I can happily report that there's no graphics corruption so far. The icons were all messed up on PS2,PS3 and some emulators. Just seems to take a little long to boot up the game.

[mineral55]

Can we report issues here? It seems the channel for reporting bugs is missing from the discord. I noticed vertical scrolling isn't in sync in Tomba! (USA) CRC32:b00ecb0d in the dwarf village when you run (holding square) up and down the screen.

This is using the latest build (25/09/2022) and the bios downloaded from the update_all script.

A workaround I found is setting the game to PAL. I've attached the savestate if anyone wants to see

Can anyone reproduce this?

[FPGAzumSpass]

It seems the game cannot hold it's 60fps in this scene, which results in tearing due to a missing frame.
Slight CPU turbo already helps.
Setting to PAL reduces the overall speed by 1%, could be enough also.

Cannot say if the real hardware also has this performance issue or not.

[mineral55]

Cannot say if the real hardware also has this performance issue or not.

It could be an edge case but I used to own the NTSC version and never encountered tearing before

[Neocaron]

The core is absolutely amazing! On a side note, I did notice that the filtering activates over sprites, in Alien Trilogy, a DBZ combat game, and a few other that I tested. Anyway to make the filtering even more selective, or maybe a bit less intense? I tried a few settings but I find the effects a bit to intense (blurry) even for 3d objects. Do you thinks it's possible to make the intensity of the filtering adjustable in some way?
24 bits is so amazing btw, bravo!

[FPGAzumSpass]

Made some comparison shots of how it would look in different steps. (first is off, last is current)

Please tell me what you think about it.

filter_intensity.png (864.2 KiB) Viewed 5634 times

[Neocaron]

Made some comparison shots of how it would look in different steps. (first is off, last is current)

Please tell me what you think about it.

I like it a lot! it looks great this way, really like 0.5 but I would add a 0.75 step as well. it would look something like "off, 0.25 , 0.50, 0.75, 1"
If this structure of tweaking is possible in the menu it's awesome. It'll allow an extra control for the filtering depending on the games or the scanline filters/shadow masks chosen. Alien trilogy would work great at 0.5, but Toy Story 3 would be more of a 0.75 or 1.

Anyway it's always awesome to see you all hears! Thank you!

[FPGAzumSpass]

Implemented now with 25,50,75 and 100%

[lsnderick]

I've been playing with filtering recently, and noticed that it's not applied to the outer edges of semi transparent textures, like the edges of the trees on the screenshots above, or the hole in the barrel in GEX2 main menu. Is this by design? Or some kind of a MiSTer hardware limitation?

[FPGAzumSpass]

It's by design and a limitation of the texture knowledge:
the gpu doesn't know the size of a texture at all, so all i can do is assume that the texture is inside the size of what the polygon itself is using.

E.g. polygon uses texture coordinates of 7..13, but the texture is in reality 0..15.
Then i cannot know it's 0..15 but only use 7..13

That also means that at the edges, i cannot interpolate with adjacent pixels in the texture, as i don't know if they are still in the texture.

[TobyRieper]

I really, really like the texturing filters. Could you expand my knowledge of the following types?
'All Polygon' is that just the filtering of 3d assets?
"Dithered' and 'Dith+Shaded'

I am playing Silent hill with Render 24 Bit: On
Dither 24 Bit for VGA (On)
Texture Filter 'Dith + Shaded' 100%
Detect 2d filter: On
It looks fantastic on my PVM

[FPGAzumSpass]

All polygon will filter all Polygons, while the setting Dithered and Dithered+Shaded will filter only polygons that are dithered( and shaded).

The use of that is mainly for some games that have e.g. 2D backgrounds or other elements that look bad when filtered, so you can leave them out.
I usually only use All polygon + 2D detect ON, that works for most games.

[TobyRieper]

Thanks for the clarification I play the alot of survival horror games on the system I will try your settings. I cant be bothered changing the settings to each game. I like the catch all approach. I noticed i get weird artifacts on Dino Crisis 2 with Render 24 bit on.

[FPGAzumSpass]

Can we report issues here? It seems the channel for reporting bugs is missing from the discord. I noticed vertical scrolling isn't in sync in Tomba! (USA) CRC32:b00ecb0d in the dwarf village when you run (holding square) up and down the screen.

This is using the latest build (25/09/2022) and the bios downloaded from the update_all script.

A workaround I found is setting the game to PAL. I've attached the savestate if anyone wants to see

Can anyone reproduce this?

retested the issue again with the latest release and it doesn't happen anymore for me.

Probably the DMA timing changes fixed it.

[mineral55]

I tested from the savestate again with the new build and I'm getting some tearing when I enter and leave the north hut still. I'm not 100% sure if it's normal behavior does anyone know?

[redsteakraw]

Just loaded up Armored Core and the text box borders are out of place, you can see everything but there is a box over the letters when you start up the input field for your name. k