PC Engine CD Support is Now On MiSTer

Galgomite · Post by **Galgomite** » Mon May 18, 2020 2:05 am

In case anyone missed it the Turbografx MiSTer core was updated to play CD games. Arcade card, fast loading and 6-button gamepads are all supported (leave it at 2-button if it's a 2-button game). It's listed as playing .bin/ .cue files but also plays .iso/ .cue so far. BTW some CD titles require Arcade card support to be toggled off to run (this was true on original hardware also). If you get a still image and can't read the Japanese message, this is the most likely problem. Pezz82 has a decent video on the topic.

Fudoh · Post by **Fudoh** » Mon May 18, 2020 8:32 am

feels and plays great. Highly recommend.

Unfortunately the option to increase the sprite limit has been removed from the core with this update, which is a shame since it was huge improvement before. Will hopefully be back in a future update.

James-F · Post by **James-F** » Mon May 18, 2020 11:05 am

The main CPU (HuC6280) was completely redone and a ton of bugs were fixed, it is now more accurate than mednafen or any other emulator.
Losing the sprite limit switch was a small price to pay.

fernan1234 · Post by **fernan1234** » Mon May 18, 2020 11:49 am

It's also already more accurate than the SSDS3 ODE. Not sure how it compares to the UGX-02 since I don't have experience with that one. Seek time related performance is also better than anything else I know of (including CD-ROM drives in good shape).

Bratwurst · Post by **Bratwurst** » Mon May 18, 2020 1:17 pm

fernan1234 wrote:It's also already more accurate than the SSDS3 ODE. Not sure how it compares to the UGX-02 since I don't have experience with that one. Seek time related performance is also better than anything else I know of (including CD-ROM drives in good shape).

Somebody turn this into an ODE I can shove in my Turbografx CD drive please.

azmun · Post by **azmun** » Tue May 19, 2020 3:19 am

fernan1234 wrote:It's also already more accurate than the SSDS3 ODE.

How is it more accurate? SSDS3 works (granted it has its flaws) with original hardware and outputs on my SDTV. Can MiSTer output 240p natively? My biggest gripe with these emulators/clone systems is their display options.

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 3:46 am

azmun wrote:How is it more accurate?

Simple, glitches and inaccuracies present when using the SSDS3 are not present on this core. Plus, unlike the SSDS3, where bug reports have been sitting unresolved for 1-2 years, bugs reported on github for this open source project are being squashed as early as the same day: https://github.com/MiSTer-devel/TurboGr ... Ter/issues (about half of the 8 currently open issues are already solved in the newest commits).

azmun wrote: Can MiSTer output 240p natively?

Yes, it can output the actual resolution(s) and refresh rates of a system via analogue output. The scaler options for digital output are also the best available, by far. Even the best RGB board on real hardware for analogue or that + OSSC for digital will never be as good.

azmun wrote:My biggest gripe with these emulators/clone systems is their display options.

MiSTer is neither of these. You'd benefit a lot from looking more into it.

Syntax · Post by **Syntax** » Tue May 19, 2020 10:06 am

Seems its ruffled TO's feathers.

Spoiler

Bassa-Bassa · Post by **Bassa-Bassa** » Tue May 19, 2020 11:15 am

How's MiSTer's low latency mode for standard USB boards these days? Have the per-core controller compatibility issues been ironed out?

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 11:47 am

Equating this platform to an emulator/RPi is either plain stupidity, ignorance, or an attempt to discredit it. With how accurate this core is already, it can be said confidently to be a viable replacement for real hardware based solutions even for the hardcore and purist. And it's only getting better from there.

Bassa-Bassa wrote:How's MiSTer's low latency mode for standard USB boards these days? Have the per-core controller compatibility issues been ironed out?

Regarding input lag, for a while there's been a fast (1ms) USB polling option, which makes even old crappy controller adapter dongles or USB controllers feel literally as responsive as the real thing. There's also the SNAC adapters (this core is compatible) to use original controllers/devices that interact directly with the core, as if it were real HW.

Analogue output on a CRT of course has no added lag. Digital scaled outputs have three configurations, including a lagless one though it may not be compatible with all monitors/TVs, while the most compatible mode adds 1 frame. The 0 lag mode has very high compatibility with flat panels though.

Per-core controller compatibility issues must be a very early problem, resolved a long time ago.

Bassa-Bassa · Post by **Bassa-Bassa** » Tue May 19, 2020 1:54 pm

Silly question, if it's a universal USB feature, why is it an "option"?

So for usage of USB adapters, the 1ms polling is the best way to go?

Thanks!

Josh128 · Post by **Josh128** » Tue May 19, 2020 2:05 pm

Fudoh wrote:feels and plays great. Highly recommend.

Unfortunately the option to increase the sprite limit has been removed from the core with this update, which is a shame since it was huge improvement before. Will hopefully be back in a future update.

Fudoh, what is the improvement of increasing the sprite limit? Reduced sprite flicker?

Syntax · Post by **Syntax** » Tue May 19, 2020 2:06 pm

Here is a rundown of original hardware vs LLAPI vs USB 1ms

NES polls for input every 16ms
|-------------------------16ms------------------------|
|---| this is 900us the time LLAPI takes
|.| this is LLAPI fast polling (not in use yet)
|| this is the real nes hardware
---|----|----| ----| ----| ----| ----| ----| ----| ----| ----| This is USB 1ms

with USB 1ms, this can happen
--------------------------|-------------------------16ms------------------------|
---|----|----| ----| ----| ----| ----| ----| ----| ----| ----| This is USB 1ms
See how it was done in the middle? That is jitter...That is bad.. that you do not want

Since LLAPI is locked to the start of a poll (or rather just before it) its always on time.

|---|-------------------------16 ms-------------|
poll

The data is ready before the poll. Every-time.

For all the console cares it could be

|------------------------------------|-------------------------16 ms-------------|
poll

as long as the poll period is under 16ms

Another way to look at it.
|-------------------------16 ms------####-|-------------------------16 ms------####--|
#### = time allotted for polling.

So compared to real hardware 800us difference between a NES controller at 100us and LLAPI at 900us.
But its inside a poll, the console will never notice it

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 2:12 pm

Bassa-Bassa wrote: Silly question, if it's a universal USB feature, why is it an "option"?

So for usage of USB adapters, the 1ms polling is the best way to go?

I'm not sure why it's not turned on by default. Most likely because it was an added feature and devs tend to avoid turning on new features without user input. Also the lead dev commented that this mode puts the USB device in an "undocumented state", though so far I haven't heard of anyone having any problems with it.
Yes, it's definitely the best way to go for anyone.

Syntax wrote:with USB 1ms, this can happen
--------------------------|-------------------------16ms------------------------|
---|----|----| ----| ----| ----| ----| ----| ----| ----| ----| This is USB 1ms
See how it was done in the middle? That is jitter...That is bad.. that you do not want

Since LLAPI is locked to the start of a poll (or rather just before it) its always on time.

Yes very technically, but unlikely to happen during a crucial input. Let's be real, you're not going to notice it. It's for practical purposes as good as LLAPI. But, for absolute OCD-level peace of mind, LLAPI options are there too. Right now the only board available is the BliSSter, which is not officially supported due to partial closed source code on it. There's an open source one in the works that will come out eventually. And again, for the few cores that currently support it, there's also SNAC. But honestly, 1ms polling should be good for pretty much anyone, except maybe cases like SMB1 speedrunners.

shroom2k · Post by **shroom2k** » Tue May 19, 2020 3:12 pm

fernan1234 wrote:I'm not sure why it's not turned on by default.

Breaks compatibility with some gamepads afaik.

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 4:12 pm

shroom2k wrote:Breaks compatibility with some gamepads afaik.

Could be, though everything I've tried so far with it (Raphnet Famicom adapter, Mayflash adapter for Saturn, N64, PS2, PS1, and SNES... Sony DS3, DS4, some brandless USB pad, forget what else...) all works fine. Anyway, you just need to set it and forget it. I heard the PC Engine mini pads work also, so that will pair nicely with this core.

azmun · Post by **azmun** » Tue May 19, 2020 4:45 pm

fernan1234 wrote:Yes, it can output the actual resolution(s) and refresh rates of a system via analogue output. The scaler options for digital output are also the best available, by far. Even the best RGB board on real hardware for analogue or that + OSSC for digital will never be as good.

I guess accurate and authentic are two different things. Not sure if I can ever get used to playing these classic consoles digitally (i.e. on flat panels). I've tried with my Mega Sg. There's a little fine point here. I did some research and sadly, it seems you can't simply run MiSTer on a consumer grade standard definition television with legacy signals (rf, composite or s-video). I know RGB is all the rave, but I've gotten back to basics and it's quite satisfying (fuzzy, dithering and all).

I figured, if I really wanted to get the MiSTer to run on my preferred display, I have two options:
a)buy either an HDMI to VGA adapter/converter OR
b)buy the I/O board

Either way, you'll need additional gear, a VGA to composite / s-video converter. Not sure if anyone has gone this route, but I've not read any reports. This last equipment may very well be the weak link?

cr4zymanz0r · Post by **cr4zymanz0r** » Tue May 19, 2020 4:49 pm

azmun wrote: I guess accurate and authentic are two different things. Not sure if I can ever get used to playing these classic consoles digitally (i.e. on flat panels). I've tried with my Mega Sg. There's a little fine point here. I did some research and sadly, it seems you can't simply run MiSTer on a consumer grade standard definition television with legacy signals (rf, composite or s-video). I know RGB is all the rave, but I've gotten back to basics and it's quite satisfying (fuzzy, dithering and all).

It has native support for component video as well. The Genesis core has a composite blending option that kinda softens things to blend dithered patterns together. It even has a dynamic blending option that blends only the dithered layers that need to be (from what I hear it's very good, but there are still some edge case games that don't handle the dynamic blending well).

hellbelly · Post by **hellbelly** » Tue May 19, 2020 4:54 pm

Here’s an open source adapter if RGB isn’t your preference, from a reliable company

https://www.retrotink.com/post/vga2ntsc-released

Update - discussion here with unmaker building some

viewtopic.php?f=6&t=66295

Deubeul · Post by **Deubeul** » Tue May 19, 2020 5:00 pm

fernan1234 wrote:It's also already more accurate than the SSDS3 ODE. Not sure how it compares to the UGX-02 since I don't have experience with that one. Seek time related performance is also better than anything else I know of (including CD-ROM drives in good shape).

To believe this accuracy affirmation I'd need to test Rayxanber 2's 4th level on the miSTer.

On real hardware, the slowdowns break the autofire of the avenue Pad6, and leave only the medium autofire speed usable on 2 buttons pads.

The game behaves exactly the same on the SSDS3.

Is it possible to use real PcE pads on MiSTer to begin with?

paulb_nl · Post by **paulb_nl** » Tue May 19, 2020 5:05 pm

Syntax wrote: with USB 1ms, this can happen
--------------------------|-------------------------16ms------------------------|
---|----|----| ----| ----| ----| ----| ----| ----| ----| ----| This is USB 1ms
See how it was done in the middle? That is jitter...That is bad.. that you do not want

This does not matter because with USB there is at most 1ms latency.

Since LLAPI is locked to the start of a poll (or rather just before it) its always on time.

|---|-------------------------16 ms-------------|
poll

The data is ready before the poll. Every-time.

This is only true if the NES polls at the same exact time every frame and it doesn't because NES games poll manually. If a game polls at scanline 50 one frame and it polls at scanline 30 the next frame then LLAPI has the input ready at scanline 50 where it should have been ready at 30 and the input is 1 frame delayed.

Genesis is one of the cores where LLAPI is synced to Vblank. If the games poll outside of Vblank then there is a much bigger delay than 1ms from USB depending on how far from Vblank it reads the controllers.

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 5:11 pm

azmun wrote:Either way, you'll need additional gear, a VGA to composite / s-video converter. Not sure if anyone has gone this route, but I've not read any reports. This last equipment may very well be the weak link?

Component for SDTVs is also an option, though that may not be authentic enough for you. S-video would not be "authentic" either for the PCE, but composite would, using a converter like the one linked above by hellbelly. Granted, we can assume that the composite converted from the MiSTer's RGB will not look exactly the same as the original composite from a PCE IFU-30, CoreGrafx, or SuperGrafx. The original composite is relatively good, so I'm actually interested in seeing this comparison once I get my hands on a VGA2NTSC.

Deubeul wrote:Is it possible to use real PcE pads on MiSTer to begin with?

Yep. Could use a raphnet adapter, those are really great quality. Or a blissbox adapter. Even a generic adapter would work. All of these will perform the same with 1ms polling.

You can do that test of Rayxanber and post how it goes. This core has 6-button support that you can enable. Though the game behaving this way when using the SSDS3 is not due to the ODE's accuracy though, since slowdown would be down to the PCE's hardware, not drive emulation. The MiSTer core is better at emulating the disc drive, which is the only thing the SSDS3 does. Bugs that exist in both, such as graphical glitches in the intro of Popful Mail, are currently much more minor on the MiSTer core than on the SSDS3, and I expect it to be fixed sooner.

edit: oh, Arcade Card emulation is also better.

Deubeul · Post by **Deubeul** » Tue May 19, 2020 5:36 pm

Would be an expensive test

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 5:45 pm

I'll try to test that for you when I get a chance. If it behaves as expected it would speak fo the CPU emulation accuracy though, not the disc drive emulation part.

edit: but really you should just get a DE10, such a great value. As amazing as this core is already, it's just the tip of the iceberg.

Bassa-Bassa · Post by **Bassa-Bassa** » Tue May 19, 2020 5:55 pm

So there're still easily noticeable glitches. I wonder how of many are there in lesser-known/Japanese-heavy games. But yeah, at least with MiSTer there're possibilities to get them fixed, unlike with the SSDS3. Anybody tried the CPS core to see if Mame's waitstates-related issues (slowdown accuracy, that is) are there as well?

fernan1234 · Post by **fernan1234** » Tue May 19, 2020 6:27 pm

Bassa-Bassa wrote:So there're still easily noticeable glitches. I wonder how of many are there in lesser-known/Japanese-heavy games. But yeah, at least with MiSTer there're possibilities to get them fixed, unlike with the SSDS3. Anybody tried the CPS core to see if Mame's waitstates-related issues (slowdown accuracy, that is) are there as well?

There are, but they seem to be very few and relatively minor so far. There's lots of knowledgeable PCE fanatics looking at it closely, including the legendary David Shadoff. The whole library including obscure Japanese titles will get well tested soon enough. And yep, one of the first things I did was grab a list of titles with issues on the SSDS3 and ran them on this core. Only one glitch was shared, and is already being looked into.

I also just realized that the newest build has the best audio balance I've heard compared to what you get from most hardware revisions I've used. Apparently it's been cross-referenced with an original PCE + Super CDROM2 with the MDFourier audio analyzer. But no audio noise from the CPU, SRAM, etc

And last I heard the CPS1 core has the slowdown accuracy, but I can't confirm personally as I'm not a Jotego patreon. The public release should be out soon though now that there's a release candidate.

Fudoh · Post by **Fudoh** » Tue May 19, 2020 6:31 pm

I also just realized that the newest build has the best audio balance I've heard compared to what you get from most hardware revisions I've used.

it's good, yes, but just like on orginal hardware effects often times outweight the music volume. Would be great to get a manual mixer to increase CDDA audio compared to chip audio. On original hardware (interface unit) you could always use the CDDA output separately from the PCE's audio output (if use with a CGfx 1 or 2) and mix the audio levels on your own. I really miss that option here on the MiSTer.

gojira54 · Post by **gojira54** » Tue May 19, 2020 7:20 pm

Deubeul wrote:
fernan1234 wrote:It's also already more accurate than the SSDS3 ODE. Not sure how it compares to the UGX-02 since I don't have experience with that one. Seek time related performance is also better than anything else I know of (including CD-ROM drives in good shape).
To believe this accuracy affirmation I'd need to test Rayxanber 2's 4th level on the miSTer.

On real hardware, the slowdowns break the autofire of the avenue Pad6, and leave only the medium autofire speed usable on 2 buttons pads.

The game behaves exactly the same on the SSDS3.

Is it possible to use real PcE pads on MiSTer to begin with?

Mister supports native pads connected directly to the core bypassing all USB interface - 'SNAC'
I can test this - will edit result later on

scrilla4rella · Post by **scrilla4rella** » Tue May 19, 2020 7:25 pm

I've been playing on this newly revised core a bunch over the last few days. It still blows my mind when I hear the CD sound coming through the speakers. Hoping they are able to implement the increased sprite limit as I was looking forward to playing R-Type Complete with it on.

Can't believe TO peeps are getting snarky over this. Makes me think twice about getting their upcoming Saturn ODE

Bassa-Bassa · Post by **Bassa-Bassa** » Tue May 19, 2020 7:34 pm

Think twice. As an old customer.

shmups.system11.org