Coleco Chameleon .... hardware speculations?

[Flojomojo]

 

Yeah, I was about to say. Needing a 3GHz PC to run 16 bit games is... not even close to being true. Hell, the Raspberry Pi 2 can run most PS games perfectly fine, which are infinitely more complex than Genesis or SNES titles.

They're quoting the specs needed by the guy who wants his emulators to be ultra-accurate. Most people (including me) are fine with "good enough."

 

We don't know what level of accuracy the Shameleon (love that!) will achieve.

[enoofu]

Wouldn't say the PS is much different then a Rasp, get different with older stuff when they didn't attempt to use standardized Processors that most desktops used and have oddball GPU solutions

[MotoRacer]

And here lies the issue.

Even byuu higan fixed the mid can line effects bug after 9Y of working on his own emulator.

The effect was just that one (yep only one) game rendered correctly the shadow under the player ship making though playing the game easier as the devs intended in the first place.

 

The general feeling is that once the various "cores" (and I mean down to the CPU, GPU, SPU etc...) are worked out FPGAs promise a more accurate emulation while using less resources, granted it is the same as saying it for SW based emus on the right PC.

But faster CPUs are no more, multi-core is the way, while FPGA has still room for improvements: hybrid systems being the proof so a generic MPU or MCU does the "management" tasks (USB, SD interface and the like) while the FPGA focuses on the simulation aspects.

 

If PC was all that was needed projects like MiST/TC64/MCC216 and others would make no sense and yet they garner quite the followers, so let's not call stupid everyone that begs to disagree with "PC based emu is all we need", it's obviously not for whatever reason. If it is for you then very good, it for sure it is not for me if I had a chance.

 

Whoa, Whoa WHOA.

 

I NEVER called anyone STUPID, first of all. I respect Leods, he's a good user here at AA. And I also NEVER claimed that computer based emulation was all we needed. Period. You twisted the words of my incredibly short post to fit your point, and I really don't appreciate it. I was simply saying a 3GHz processor isn't needed to run a Genesis or SNES game. Yikes.

Edited February 20, 2016 by MotoRacer

[Osgeld]

things like mist exist cause people want them to, FPGA only has an advantage of being able to dedicate all its limited resources to doing one task rather than some of its huge resources to multiple tasks, in the end thinking just cause it has a magic acronym makes it dead nuts on to the real metal is foolish, in both worlds it boils down to the exact same thing... the software written

 

you can have some extremely accurate PC emulation and you can have some extremely hap hazard FPGA implementation, doesnt matter if its written on a X86 or a Cyclone, its either good or its not

Edited February 20, 2016 by Osgeld

[leods]

No idea where the whole " so let's not call stupid everyone that begs to disagree with "PC based emu is all we Need""

 

Literally, no one said anything even Close to that.

 

But on the Topic at Hand, it doesn't matter what Kind of tech the CC uses, it's not automatically guarantee things will work well. Specially since they're promissing you'll use your own USB Controllers. For this Project to make any sense at all they'd Need a production ready prototype, running all games with not only their Controller, but with generic usb Controllers working too. All from cartridges.

 

One Thing I saw no one mention, Mike actually said they could use Masked ROMs for the smaller games. First time I see ROMs being brought up in a Long time. But you know, "we could use", when mike says it, pretty much means that. They could. Doesn't necessarily mean they will...

[Akito01]

http://www.giantbomb.com/podcasts/the-giant-beastcast-episode-39/1600-1510/

 

Relevant talk begins at 1:14:48

 

I don't think this has been posted yet, but in the spirit of passing along mainstream gaming press coverage of this, the most recent episode of the Beastcast does a good chunk on it. Jeff Bakalar of CNET went to the toy fair and interviewed Mike. Their impressions are super interesting and entertaining.

 

Might be worth passing along a word to beastcast@giantbomb.com ?

[+madman]

I hope this doesn't get anyone into any trouble, but an insider on the CC team leaked this photo showing early development of the Genesis core.

 

[TheClassyGamer]

I'll try my hand at this.

 

 

Also, I hear the next KS level there is going to be a Collectors Bundle that comes VGA graded.

Edited February 20, 2016 by TheClassyGamer

[mickcris]

I hope this doesn't get anyone into any trouble, but an insider on the CC team leaked this photo showing early development of the Genesis core.

 

Interesting. I figured their core would be a Genesis 3 not a 2.

[Raticon]

I don't get what they are saying about controllers anymore. First they told us that it only would take their own controller, then they said they would "look into" possibilities to use other USB controllers and all of a sudden their USB cv/inty controllers are announced and now it looks like any USB controllers can be used.

 

Do they realize they will be trashing their own controller sales now? Everyone got to have a box full of USB controllers in the house, why buy their modded Wii-U controller when i can use any other i might have around?

 

Since everything else will be proprietary on this thing this seems like an odd decision to say the least.

[Keatah]

Multiple cores actually makes emulation more difficult, especially if you're trying to get close to being cycle accurate. Keeping all the threads in sync is quite a hassle and can cause more issues than they're worth. That's why many emulators only make use of a single core.

 

I feel that not enough programmers are fluent in parallel processing. Whereas when "programming" with HDL you naturally think in parallel.

[Keatah]

I'll try my hand at this.

 

 

Also, I hear the next KS level there is going to be a Collectors Bundle that comes VGA graded.

 

 

Just call it the Collect-O-Vision.

[cybercylon]

 

 

 

What? Piko is a member here and he's been very straight forward, he clarified what concerns he is keeping to himself, the fact that the board he had at the ToyFair was actually for something else (he mentioned lootcrate).

So I see no evidence of them being in any way in on it. Not sure what gave you that impression.

 

 

The original poster (meant include quote, but hit hit the button before I could) may need to go back many pages. Easy to miss some posts when this thread is so long, but as mentioned, they have concerns, but aren't able to say much until things go more public. Are they the ones that had a contract to make a game back when this was the RetroVGS? I believe so. So certain events may need to play out before they say much, but don't see any evidence of fishiness from PikoInteractive. Hold tight as I am sure they will have more to say.

 

*Edited because I am still getting used to my new mechanical keyboard.*

Edited February 20, 2016 by cybercylon

[King_Salamon]

Interesting. I figured their core would be a Genesis 3 not a 2.

 

huh... and here I figured they were going to need more tape.

[phoenixdownita]

 

Whoa, Whoa WHOA.

 

I NEVER called anyone STUPID, first of all. I respect Leods, he's a good user here at AA. And I also NEVER claimed that computer based emulation was all we needed. Period. You twisted the words of my incredibly short post to fit your point, and I really don't appreciate it. I was simply saying a 3GHz processor isn't needed to run a Genesis or SNES game. Yikes.

Apologies, I chose the wrong quote although the emphasis was on "most games", must have been then umpteenth post about using emu that shorted my fuse. Not your fault.

 

I do use my PC for Mame, I do have an RPi2 with RetroPie and I do have most of the actual consoles I care for with related flash carts.

What I hate of the emus is all the fidgeting even to get started ... I know how to do it I just don't want to do it anymore (I admit in recent years it got better but still a far cry from where they should be) ... likely not a fault of the emus but of the platform in general.

With the actual consoles what I hate is the lack of HD out and in some notable cases the dreaded std controllers (7800, SMS among the worst offender).

 

Again no offense meant I got tired of "emu here emu there", and if the CC turns out to be just an emu with custom front end a la NeoGeo-X it would be as useless as that one.

Edited February 20, 2016 by phoenixdownita

[PlaysWithWolves]

I have no insight into Piko Interactive, but it is a company I will not buy anything from. Not after watching how

they engage in this scam. It is very off-putting to me. However, they seem to be getting off easy...maybe they're

some sort of celebs?

 

/Nicholas

 

He won a lot of points here in the RVGS debacle. He showed us how not only was the RVGS nonexistent on the hardware front, but even on the software side. When a game producer says that he has no idea how a game will get ported to a system its promised-for, people take note. It would be like the Chameleon hardware guy stepping up and saying, "We have no real prototype."

 

Is guilt-by-association fair? Beats me and I completely understand if you feel that way. Piko/Eli seems to be taking more of a wait-and-see attitude about the Chameleon. Contrast with CollectorVision (as mentioned earlier), where they produced half-hour infomercials on RVGS where it was only briefly mentioned toward the end of the video that the pack-in game was theirs. That seems a less forgivable offense to me.

[goldenegg]

I have no insight into Piko Interactive, but it is a company I will not buy anything from. Not after watching how

they engage in this scam. It is very off-putting to me. However, they seem to be getting off easy...maybe they're

some sort of celebs?

 

/Nicholas

 

 

What? Piko has absolutely nothing to do with how Mike & Co are handling the Chameleon. They're simply a developer who are looking to release some of their games on the system. They've done nothing to defend Mike and have even had some criticisms. The fact they'd rather not completely open up regarding all their concerns prior to the Kickstarter is a reasonable thing to do. Not really sure what your issue is with Piko.

[Chocobro]

I'm a hardware/FPGA programmer myself(it's what got me interested in the concept of the RVGS and led me eventually here).

 

The biggest benefit of FPGA is accuracy, resources, and access to original hardware.

 

Another post is correct that most FPGA reverse engineering is from a black box perspective, this method often can lead to 100% compatibility with enough time and effort, something software emulation can never realistically achieve.

 

That said, given access to the information(like if we XRAY'd old chips and generated chip schemas from them), it's a possibility to actually re-create the original chip, gate for gate(with some caveats depending on the FPGA chip itself). This method is not easy, but is possible, the more complex the chip becomes the more tedious and impractical it is. This method would work fairly well on 4/8bit era chips but the 16bit era chips kinda straddle the line of having too many transistors to be feasible to do 1 for 1 copies, so blackbox reverse engineering is more practical here.

 

The accuracy of the FPGA means a few other things.

 

• Compatibility with original hardware, i.e. carts/discs.
• Compatibility with original hardware accessories. A "compatible enough" FPGA chip should maintain compatibility with hardware like lightguns, game genies, etc. Hypothetically if you had a 100% compatible genesis core you could connect it to a SEGACD or a 32X.
• FPGAs could not only reach hypothetical 100% compatibility, but add new features to old machines that you'd have to mod into, or buy expensive aftermarket items for.

 

IMHO FPGAs are relatively new to the retrogaming scene, and it may take a long time(like decades) for them to be worked on, checked, made 100% compatible, but once they are, they are absolutely the best way to play games, possibly even eclipsing original hardware. It's important to remember that FPGAs aren't emulation,

it's a technology that allows you to programmably modify physical hardware, you are literally re-creating the original chips(or a variation there-of), with the prime benefit being that you can switch what the chip is, by changing its programming(swapping "cores").

 

All that said, software emulation is in most cases "good enough" for most games and most people. But in a long-term preservationist/purist sense, we should definitely be embracing FPGAs.

[Reaperman]

So I've been sniffing around looking at FPGA machines currently around, and am hugely interested in owning one.

The very reason I joined atariage way back, was because I met some AA members on the forum of a handheld fpga project.

 

Getting back to the matter at hand with fpga vs software emulation I think we might need to step back for a minute and look at fpga vs asic first.

 

As I understand it, an FPGA will perform pretty similarly to the many devices that use custom ASICs (for example nes-on-a-chips etc) with all the warts of those, but the ability to reconfigure them to do other systems or to fix issues after release. Because of that, I think we already have a good idea how an FPGA will perform, especially a system with a selling point of 'no patching.' We have very mature ASIC systems now. They're pretty awesome.

 

I could be wrong on all of this, so please correct me. Honestly this is a FAR more interesting topic to me than some peddler with wild dreams. An FPGA system without a strong and continuing development process has no significant advantages over much less expensive asic systems aside from maybe handling some additional oddball systems.

Edited February 20, 2016 by Reaperman

[Keatah]

Dumb Question time;

 

Okay, so it looks like everyone's just going to be rehashing the same lack of news from now until the kickstarter, I thought it might be a good time to ask a dumb question. Maybe I am not the only one wondering about this…

 

At the risk of sounding trollish, I want to ask -

 

What's the big deal with FPGA? Why is it considered the "gold standard" for emulation?

 

A lot of folks believe FPGA recreates the hardware exactly - resulting in perfect emulation. This is not true at all. There is no transistor-level recreation going on.

 

FPGA sometimes gets the "gold-standard" label is because the simulation runs without interruption from a host OS. Early software emulation was always stuttering and hiccuping due to an underpowered processor that also had to perform host duties as well. Software emulation is still trying to shake that bad rap.

 

This is no longer true with 3GHz machines being entry-level. More than enough power to do it all.

 

 

If we had a gate level specification of, say, the SNES video chip, then I guess anyone could implement it with gate arrays, and we would be confident that the implementation would be accurate.

 

But, we don't have a gate level spec. The SNES video chip is a "black box" to us - all we have are the results of people's experiments ("when I write to this address, this happened)" and people's ideas about what is going on inside.

 

 

That's right. As decapping and reverse engineering and other detective work improves and continues, the knowledge of the black box behavior goes up. Emulations get better, both hardware and software varieties.

 

 

So. the FPGA programmer has to try and recreate the design based on this incomplete knowledge. I imagine the quality and accuracy of the result is more dependent on the experience and skill of the programmer than anything else. There is nothing magical about using a gate array that makes this redesigned chip automatically more accurate than a software emulation of that chip, is there?

 

That's correct. It's like airplanes and helicopters. Only as good as their manufacturers, materials, and pilots. Both obey the laws of physics and aerodynamics, and "assemble" them in different ways to achieve the same end goal - fly your sagging ass from here to there.

 

 

 

It seems to me that software emulation is :

1. cheaper.

2. understood and supported by more people than gate array programming.

3. Easier to update.

 

Correct on all three points.

 

1- all you need is a contemporary PC. You probably have one already. And there are other platforms you can use instead of the PC, too.

2- this is because the hardware is more niche and not as pervasive as the cheap stuff you can get at wal-mart.

3- yup. Just download from the developer's homepage and install over your existing setup.

 

 

Why should I want a FPGA?

 

 

At this present time, you probably don't. Not until a feature rich rig comes along.

 

With a modern-day host PC you get so much more in the amenities & extras department than a stand-alone FPGA box can offer. You get screenshots, save states, over/underclocking, support for many different monitors and the ability to render simulated CRT effects with lots of user-adjustable settings, easy management and storage of collections of roms and disk images like in Stella, easy to set-up and configure console firmware/bios like in Altirra, and many many other things.

 

While nothing prevents those things from being done on an FPGA emulation rig, it has yet to be seen.

 

Early FPGA arcade simulations had just the main chip and nothing else, they ran only one or two games and that was it. Now we're seeing ARM SOC + FPGA together. Yet there needs to be significant front-end improvement because current examples don't offer feature-rich goodies like Altirra or MAME.

 

To simply run an Atari 400/800 core with 2 joystick ports is not enough, you need keyboard and controller mapping, easy management, and peripheral emulation, SIO support, adjustable artifacting and palette control, support for different firmware and memory configurations. And more. Software emulation can provide that today - right now.

[Chocobro]

Yeah, FPGAs will perform very comparability to ASICs, FPGAs are what are typically used to make ASICs. You get a working design on an FPGA, and then you get the final version made into dedicated chips(ASIC) because it's much cheaper than the FPGA chip(the FPGA chip might be $100 each but the ASIC might be $1/each). Though this isn't always true, for example it's common to see FPGA chips in really expensive devices like oscilloscopes these days, where paying extra for the FPGA chip is cheaper than taping out an ASIC because they won't sell enough volume(you need to make a lot of chips to get costs down).

 

But they perform the same, the big difference is cost and gate count/transistor count. FPGAs are much more expensive per gate, and this is the trade-off of their flexibility to reprogram them. That's what currently is the struggle with projects like RVGS and I imagine kevtris's board, in that getting an FPGA chip on the board that does everything we'd want can be prohibitively expensive, so you gotta cut features or limit platforms one way or another.

 

10yrs from now when the technology has advanced, $100 will probably buy you an FPGA that had the gate count to emulate 32/64era platforms well, where as right now $100~ is on the fence of being able to accommodate 16bit era platforms. Bigger ones are available, they just might run into the thousands of dollars and are really meant for companies doing ASIC design.

[Keatah]

So I've been sniffing around looking at FPGA machines currently around, and am hugely interested in owning one.

The very reason I joined atariage way back, was because I met some AA members on the forum of a handheld fpga project.

 

Getting back to the matter at hand with fpga vs software emulation I think we might need to step back for a minute and look at fpga vs asic first.

 

As I understand it, an FPGA will perform pretty similarly to the many devices that use custom ASICs (for example nes-on-a-chips etc) with all the warts of those, but the ability to reconfigure them to do other systems or to fix issues. Because of that, I think we already have a good idea how an FPGA will perform, especially a system with a selling point of 'no patching.' We have very mature ASIC systems now. They're pretty awesome.

 

FPGA and ASIC can be considered one and the same in this context. In fact, many ASICs are designed and tested on FPGA development boards first. In some cases it is best to not even bother with the final ASIC stage. Your hardware becomes field upgradeable.

 

"Patching" a complex system is a given and a requirement in this day and age, what with all the features and functionality and new things. More a requirement now then ever!

 

I could be wrong on all of this, so please correct me. Honestly this is a FAR more interesting topic to me than some peddler with wild dreams. An FPGA system without a strong and continuing development process has no significant advantages over much less expensive asic systems aside from maybe handling some additional oddball systems.

 

True enough, any emulator benefits from continued development. FPGA rigs are no exception.

 

This is certainly more enlightening and interesting than some dumb-ass vaporware product! No doubt.

Edited February 20, 2016 by Keatah

[Gentlegamer]

I dunno how much better this day can get, Ian stops by and i see my favorite user bring back my favorite phrase

 

[Keatah]

Yeah, FPGAs will perform very comparability to ASICs, FPGAs are what are typically used to make ASICs. You get a working design on an FPGA, and then you get the final version made into dedicated chips(ASIC) because it's much cheaper than the FPGA chip(the FPGA chip might be $100 each but the ASIC might be $1/each). Though this isn't always true, for example it's common to see FPGA chips in really expensive devices like oscilloscopes these days, where paying extra for the FPGA chip is cheaper than taping out an ASIC because they won't sell enough volume(you need to make a lot of chips to get costs down).

 

But they perform the same, the big difference is cost and gate count/transistor count. FPGAs are much more expensive per gate, and this is the trade-off of their flexibility to reprogram them. That's what currently is the struggle with projects like RVGS and I imagine kevtris's board, in that getting an FPGA chip on the board that does everything we'd want can be prohibitively expensive, so you gotta cut features or limit platforms one way or another.

 

10yrs from now when the technology has advanced, $100 will probably buy you an FPGA that had the gate count to emulate 32/64era platforms well, where as right now $100~ is on the fence of being able to accommodate 16bit era platforms. Bigger ones are available, they just might run into the thousands of dollars and are really meant for companies doing ASIC design.

 

Agree on the cost/performance summary.

 

There are FPGA chips that run $10,000 and upwards.

Check the Stratix lineup.

 

And then you have intel wanting to put fpga on their processors..

http://www.extremetech.com/extreme/184828-intel-unveils-new-xeon-chip-with-integrated-fpga-touts-20x-performance-boost

[CPUWIZ]

Random thought, why does anyone care about how much an item that entertains you for hours on end costs? I mean this forum is full of people who will pay several thousand dollars for a single boxed game, just to shove it onto a shelve, why cry over a $100 FPGA?