What would you buy?

[applekevin]

But what if this emulation was *exactly* like the original? What if the game play was the same? It is possible to emulate and have the same old Atari feel! Before you say that the FB1 emulated and it sucked, you're wrong because it didn't emulate it simulated and there is a huge difference. A simulation is a rewrite an emulation is not. There is good and bad emulation. Does anybody own a MAME box? A mame box is an arcade machine that emulates arcade hardware to allow you to play classic arcade games. Now if you want to play arcade games are you going to emulate it or rebuild the arcade hardware. Obviously you would emulate it. Here are a few reasons why: 1. Most of the hardware used to play classic arcade games isn't manufactured anymore or isn't available to the public. 2. Different arcade games use different hardware, if you wan to play 20 different games you can't have twenty arcade cabinents in your house. This applies also to the FB3. If you're using the same joysticks, the *exact* same games, the *exact* same look and feel of the case what’s the difference? Also if it's compatible with all 2600 7800 5200 and A8 games what would you buy? An atari-on-a-chip that's 80% compatible with only 2600 games or a emulated system that is 100% compatible with ALL classic Atari games? If it looks the same plays the same and feels the same what's the difference? You guys want 100% compatibilty. There are also a lot of people who want a 2600/7800/5200 omniatari (such as my self!) I'll tell you right now that will NEVER happen. 1st the hardware isn't made anymore that was used in the classic Atari systems so you would have to recreate it which will never have 100% compatibility. 2nd it would be way too expensive to make a 2600/5200/7800 because of the difference in hardware. Atari will NEVER do that because most people won't know the difference.

 

OK I'll stop rambling now! :-)

 

P.S. I have a MAME box that I built and to tell you guys the truth THERE IS NO DIFFERENCE IN THE REAL THING! I mean I can’t tell at all! Plus I can play almost any arcade game I want without having to buy more hardware. Just consider: Do you want real Atari hardware with 80% compatibility with ONLY the 2600 –OR—Emulated Atari hardware compatible 100% with 2600/5200/7800/A8 games? It could even include cart ports in theory! Just a thought

[Feralstorm]

And how much did it cost you to build that MAME box?

 

The problem with emulation is it generally requires more hefty computing hardware than the 20-50 dollar price tag of most TV home games would allow. That's why they pretty much all use new versions of old hardware with the original code (FB2, C64-DTV), or some other off-the-shelf components with ported version of games - it's way cheaper.

Edited December 24, 2006 by Feralstorm

[applekevin]

Actually A mame box doesn't cost me that much- for the computer! I picked up a $20 computer for it. Reproducing an old chip for emulation wouldn't cost very much. The 2600 was only about 1.5 Mhz!

[nathanallan]

So Curt, is this thread helping any?

 

Nathan

[mstulir]

So Curt, is this thread helping any?

 

Nathan

 

Nathan, Curt is on vacation with his wife and daughter until after Jan. 1, so you won't get an answer from him for another week.

 

Believe me, the team is watching. We are listening to what people are saying and making sure it gets passed up the ladder. We want to make the best device we can, and all of the opinions and encouragement here is a big help. We do not have any news we can make public yet, but as soon as we can inform the masses, Curt will post it here first.

[mstulir]

Reproducing an old chip for emulation wouldn't cost very much.

 

And you base this assumption on what?

[Flojomojo]

Actually A mame box doesn't cost me that much- for the computer! I picked up a $20 computer for it. Reproducing an old chip for emulation wouldn't cost very much. The 2600 was only about 1.5 Mhz!

 

 

Reproducing an old chip for emulation wouldn't cost very much.

 

And you base this assumption on what?

From the flea market PC he got for $20.

 

Kevin, manufacturing is more expensive than scrounging. It's not like they're cracking open old Ballblazer cartridges to steal POKEY chips here. Just because something is old and slow doesn't necessarily mean it's cheap -- look at the Battlesphere saga on Jaguar for a lesson in scarcity of vintage parts. Another example: it's often cheaper to buy RAM for current systems than for computers that have been out of production for a while.

 

There's no reason to use the old chip from the VCS when a new design could use less power and take up less space, especially for use in a portable. That means designers and manufacturers ... which by definition means massive quantities since it's cheaper to make these things if you can order a lot of them.

[mstulir]

From the flea market PC he got for $20.

 

Damn, I was really hoping that kid knew something I didn't. I was looking so forward to the technical explanation from "applekevin" as to how we could have designed a $9 board to emulate a 2600 and that all of the engineering work Curt, Marty & myself have done in the last 3 years was wrong.

 

AppleKevin, I don't want to burst your bubble or destroy a fantasy, but let's make sure you understand something: There is NO SUCH THING as 100% accurate emulation. Your earlier statement about there being no difference with the real thing is incorrect. If you do not believe me, I suggest you attend a gaming tournament with championship-caliber players (such as the Funspot tournament) and listen to those guys discuss the flaws in MAME vs. the real hardware. I learn more from those guys than I ever would have expected. That was one of the reasons I took a pre-production Flashback 2 up to the Funspot tournament in May of 2005 -- I knew that if there was a serious flaw in our engineering work, those guys would find it.

 

MAME is pretty good, but the programming team have always made it clear that they cannot achieve 100% perfection. Their goal is to document and preserve the games -- not reproduce them EXACTLY as they were. That statement is on the Index page at the main MAME web site.

 

Flojomojo is absolutely correct when he mentioned the costs involved in manufacturing. AppleKevin, you can't just build a board to emulate all of those systems to go into a $29 device. In the Omniatari thread, you said that you would be willing to pay $60-$80 for such a device. I have news for you....it wouldn't be that cheap.

Edited December 27, 2006 by mstulir

[Retro Rogue]

Nathan, Curt is on vacation with his wife and daughter until after Jan. 1, so you won't get an answer from him for another week.

 

Sssshhhhh, ixnay on the acationvay. Curt's not allowed to be on vacation, he's got to be in his Atari bat cave 24/7.

[Paranoid]

From the flea market PC he got for $20.

 

Damn, I was really hoping that kid knew something I didn't. I was looking so forward to the technical explanation from "applekevin" as to how we could have designed a $9 board to emulate a 2600 and that all of the engineering work Curt, Marty & myself have done in the last 3 years was wrong.

 

AppleKevin, I don't want to burst your bubble or destroy a fantasy, but let's make sure you understand something: There is NO SUCH THING as 100% accurate emulation. Your earlier statement about there being no difference with the real thing is incorrect. If you do not believe me, I suggest you attend a gaming tournament with championship-caliber players (such as the Funspot tournament) and listen to those guys discuss the flaws in MAME vs. the real hardware. I learn more from those guys than I ever would have expected. That was one of the reasons I took a pre-production Flashback 2 up to the Funspot tournament in May of 2005 -- I knew that if there was a serious flaw in our engineering work, those guys would find it.

 

MAME is pretty good, but the programming team have always made it clear that they cannot achieve 100% perfection. Their goal is to document and preserve the games -- not reproduce them EXACTLY as they were. That statement is on the Index page at the main MAME web site.

 

Flojomojo is absolutely correct when he mentioned the costs involved in manufacturing. AppleKevin, you can't just build a board to emulate all of those systems to go into a $29 device. In the Omniatari thread, you said that you would be willing to pay $60-$80 for such a device. I have news for you....it wouldn't be that cheap.

 

But, you don't have to please the incredibly anally retentive purists to make a mass-consumer device that will be widely accepted.

 

I see the counterargument here. My MAME box cost me all of... probably $200 and my time. Free 1.7ghz PC, extra monitor I had lying around, free cabinet... some paint and an X-Arcade control panel and my time. But, that model doesn't move well to a mass produced item, obviously.

 

The fact that it cost me that much *and* I cut so many cost-corners really speaks volumes about the difficulty involved, too.

 

But, I also see that a modern chipset that emulates more or less accurately the original platform and will run the original code, might be less expensive than recreating the whole thing.

 

I guess that is where I'm going with this. I don't care HOW Atari does it... as long as they do it, and do it more or less acceptably. I think the bar has been set. We expect something of the same quality as the FB2. No matter what you do... there is going to be a minority that bitches about it. I could care less about them, and I don't think Atari should worry too much about them, either.

 

But to satisfy the vast majority of consumers... you want something better than *decent*. However you achieve that so that you can be profitable, is really your call, IMHO. Atari needs to make money off the deal, and you guys need to get paid. That seems reasonable enough. You guys aren't doing this for purely alturistic motivations. If you have to make some concessions to get to that point, so be it. But there is a certain point where you should go, "We just shouldn't do this, because it is going to turn out as such a steaming piece of crap, we'll only earn ill-will among the community".

 

You guys love the stuff as much as us. You'll know where that point is.

[mstulir]

Sssshhhhh, ixnay on the acationvay. Curt's not allowed to be on vacation, he's got to be in his Atari bat cave 24/7.

 

I think Mrs. Batman would be very upset if the Atari Caped Crusader didn't come out of the batcave on occasion. Actually, considering his office is currently in the basement and looking at all of the gadgets he has around his office, the "batcave" terminology is quite appropriate. All he needs is a pole to slide down so he can enter the cave from the room above.

 

I'm suddenly having visions of Curt in a Batman outfit with a fuji across his chest instead of the bat logo.

[mstulir]

But, I also see that a modern chipset that emulates more or less accurately the original platform and will run the original code, might be less expensive than recreating the whole thing.

 

I do not know where you are getting that idea. I could not disagree with this more.

 

Let me give you an example... Curt and I were involved in a non-Atari project in late 2004/early 2005 that never saw the light of day. We needed a board to perform some emulation, and Curt sat down with his tools and designed the board to do it. I can tell you right now that you cannot put together a board for console emulation that will fit into the price range that the average consumer is willing to pay. It has been a while and it was a bad experience I have tried to block out of my mind, but I believe the boards were going to cost around $100 each to manufacture. Now, do the math... By the time all of the other development costs, manufacturing, packaging, shipping, and retail markup gets added in, you are probably looking at somewhere in the neighborhood of $225-$250 for such a device.

 

Paranoid, you are absolutely correct when you say that the bar was set high with FB2. You are also correct that we do not have to please the incredibly anally retentive purists to make a mass-consumer device that will be widely accepted. The units that Jakks Pacific continues to sell is proof of that. Why aren't they emulating in their Namco & Pac-Man units? It is probably because it is not cost effective for the average consumer.

Edited December 28, 2006 by mstulir

[jbanes]

Let me give you an example... Curt and I were involved in a non-Atari project in late 2004/early 2005 that never saw the light of day. We needed a board to perform some emulation, and Curt sat down with his tools and designed the board to do it. I can tell you right now that you cannot put together a board for console emulation that will fit into the price range that the average consumer is willing to pay. It has been a while and it was a bad experience I have tried to block out of my mind, but I believe the boards were going to cost around $100 each to manufacture. Now, do the math... By the time all of the other development costs, manufacturing, packaging, shipping, and retail markup gets added in, you are probably looking at somewhere in the neighborhood of $225-$250 for such a device.

I'm going to back this up. One of my first ideas for a retro, home game console was a MAME emulator with pluggable machine modules to support new games. I figured that it would be easy to grab an ARM processor and development board and get to work. Unfortunately, I found that such a path was way too expensive for the toy market. ARM cores and comparable microprocessors are only slightly less expensive than the latest desktop chips. Plus their complexity drives up the price of the mainboard. In the end, it just wasn't practical to produce a generic board.

 

What I found was that genericity had its price. The only way that electronic toys get to market is by stripping the system down to the absolute essentials for the target product, and throwing away the rest. This design philosophy is why game consoles never made very good home computers. Their game-focused design left them unequipped and underpowered for the additional I/O and peripheral support necessary.

 

Best to leave the emulation market to the 360 and Wii.

Edited December 28, 2006 by jbanes

[Paranoid]

Well... alright then.

 

How *are* the Jakks-Offs doing Ms. Pac Man, Xevious, Mappy, etc. so faithfully? I mean, those games are close enough to the originals that I would have guessed they were emulation running the orignal code. Are they actually PORTS of the original code?

 

I guess that may be the problem. It just seems that there should be low end, off the shelf processors available *cheap* that should be able to do a decent job of emulation on something like say, a 5200 or 8 bit.

 

There is nothing between a real low end processor/chipset suitable for a PnP toy and something MORE powerful like an ARM processor suitable for a PDA? Even with economies of scale?

 

Seems like a niche that needs to be filled, at the very least.

 

I'm thinking of they can do those Jakks sticks at what... $29 initial retail... and a PDA can be found as low as $200 retail... there should be something that allows you to come in at a retail price somewhere between those two, while delivering a quality product. And my gut intuition says that if the low end PDA at $200 can do a decent job with retro emulation... this hypothetical product should be able to, also.

 

Not that I'm arguing with either of you. Clearly you've both been farther down this path than I have, and your statements are fact, where mine aren't even opinions (maybe more like wishful thinking, evidently).

 

Anyhow, if you *can't* do emulation, I think that is a major obstacle to the FB3 that would require you to stick to original hardware. In my opinion, the only way for this to be -largely- commercially viable is for it to execute *original* code.

 

And that might be the way to present it. If you go with new hardware that *mimics* the original hardware and runs look-alikes *or* ports of the original code, in either case, you're going to have a lot more production costs associated with converting or recreating those programs, and you'll *never* have the full library available (for whatever platform you go with), simply because of licensing issues. If your hardware will run the original code, you're going to save a bundle on packaging your *own* titles (just dump the ROM images onto whatever card factor you decide to go with, more or less, and retail package it. I know this is something of an oversimplification, but you get my point), but you also leave open an easy way for um... "whoever", to get the REST of those titles in the 3rd party libraries over and running on the device... and that is probably a great way to encourage sales. This is really why I want to see the FB3 have a non-proprietary memory card interface. If it is some patented, proprietary game-key type memory slot... that is going to stifle support for the device. If it is something with a low barrier to entry, the odds are it'll gain wider support, either through legitimate channels or... otherwise. It would also help if you can convince management that they shouldn't spend anything worrying about DRM or copy protection for 30 year old titles that are already flooded across the net. It is another thing that is just going to add expense... and if you leverage that right, this "underground" is going to actually *increase* hardware sales. You might lose a little here and there on potential software sales, but you'll more than make up for it in other ways, simply through increased proliferation of your actual hardware, at the very least. It goes hand in hand that "IBM" PCs are so successful and Microsoft OSes are the most pirated software in the world, for example. You'll still *get* software sales, if you want to make packages available, and your losses here will be marginal, because those who will "pirate" the titles, already know that they can do it without your hardware, anyhow.

 

Otherwise, you should just stick to releasing Flasback 2 v.5, v.61, etc... with a new bundle of 40 games on each one.

Edited December 28, 2006 by Paranoid

[mstulir]

 

Oh also it is much easier to reach 100% compatibilty with emulation rather than recreating the system.

 

SOme people are saying emus are uneconomical. Couldn't you use a 30 Mhz Moto 68000 to do that? DO they stil make those?

 

I'm still waiting for AppleKevin to provide the technical details on these pearls of wisdom. I think I will be waiting a while.

[Flojomojo]

Easy now ... some people are more technical than others. Some people know just enough to be dangerous!

 

I agree that it "seems" that a sufficiently-powerful-for-emulation CPU should be out there that can be mass-produced in quantity, but it's probably not the case.

 

Low-powered, ported games are fine alternatives *if* the ports are good. The Atari-on-a-paddle game is excellent. Jakks' arcade ports are good. The Atari-VCS-in-a-joystick is just OK. The Atari Flashback (1) and Intellivision plug-n-plays smell like ass.

 

I'm repeating myself and everyone else in the thread: re-engineered chips that simulate the real hardware and run the original software, such as the Flashback 2 and Commodore 64 30-in-1, are the best we've got right now.

[jbanes]

How *are* the Jakks-Offs doing Ms. Pac Man, Xevious, Mappy, etc. so faithfully? I mean, those games are close enough to the originals that I would have guessed they were emulation running the orignal code. Are they actually PORTS of the original code?

Ms. Pac Man, Xevious, Galaga, Mappy, Dig Dug, and Pole Position all run on the same hardware. Plus or minus a few custom ICs for starfield generation. (I know because I went mucking with CottAGE source code at one point to see if I could reproduce the DigDug music in a 4K game.) My best guess to date is that the Ms. Pac Man system uses a custom ASIC that combines the Galaga/Pacman platform into a System On a Chip. Which would explain how they're so accurate to the originals. (Because they are.) If anyone knows differently, please speak up.

 

There is nothing between a real low end processor/chipset suitable for a PnP toy and something MORE powerful like an ARM processor suitable for a PDA? Even with economies of scale?

FPGA simulations. It's doubtful you could do it at a $30 price point, but you could probably hit $50-$70 depending on how large of a fabric you need. You could then distribute games as a combination of an FPGA design and ROM file. Once inserted, the FPGA will be loaded with a new SoC and begin executing the ROM image. The size of the FPGA would determine what the cutoff is for complexity of arcade games. e.g. You might size the fabric to be able to simulate older games like Dig Dug, but not newer games like R-Type.

 

Keep in mind that you have to keep the FPGA size reasonable. As you attempt larger SoCs, you're eventually going to get to an FPGA price point that hits the low-end of what an ARM system would cost.

 

There are several caveats to keep in mind with this:

1. The FPGA design file can easily hit a megabyte or more. This will increase the cost of your media.
   
2. You'd need to put as much directly on the chip as possible. The more you move off the FPGA, the more difficult it's going to be to design an SoC that works as the ROM expects it to.
   
3. For ultimate compatibility, you'll need a rasterizer on your SoC to render to a framebuffer. This will eliminate the signal differences and make all the games compatible with a regular television.
   
4. Different platforms run at different clock speeds. You're going to need either a programmable oscillator, or subdivide a high speed oscillator to approximately match the speed of the original equipment. It's probably not going to be exact, but it could be close.
   
5. There's only so much memory you can keep on the FPGA before the costs become prohibitive. If you need a lot of RAM, you're going to need an external chip. That chip will probably have different timings than the original system memory that need to be accounted for.
   

Of course, I'm just brainstorming here. The practicality of such a maneuver is difficult to gauge. There are tons of potential hidden costs, especially in designing new SoCs to support a given arcade platform.

 

If anyone has some thoughts to add, feel free to jump in.

 

[/end brainstorm session]

[mstulir]

How *are* the Jakks-Offs doing Ms. Pac Man, Xevious, Mappy, etc. so faithfully? I mean, those games are close enough to the originals that I would have guessed they were emulation running the orignal code. Are they actually PORTS of the original code?

 

Admittedly, I may be looking at the Jakks product line from a different point of view than the average consumer. I know in the case of the Atari 10-In-1 joystick they did (and assuming on the other games), the games were ported. I do not want to sound like I am against porting. I think it has its place if it is done well.

 

 

I'm thinking of they can do those Jakks sticks at what... $29 initial retail... and a PDA can be found as low as $200 retail... there should be something that allows you to come in at a retail price somewhere between those two, while delivering a quality product. And my gut intuition says that if the low end PDA at $200 can do a decent job with retro emulation... this hypothetical product should be able to, also.

 

It isn't quite that simple. I think Jakks is dealing with very inexpensive parts and banking on selling the games in volume due to the low price.

 

As for the handheld PDA, I think you are not counting the costs involved for whatever custom code has to be written to run the games. It isn't just the costs of the parts. Look at the GP2X for example. It is a $200 device that does emulation quite well. There are minimal costs associated with developing the OS as it runs a derivative of Linux. As for the software, the GP2X community has done the coding themselves and it was not left up to the Gamepark Holdings crew to hire/pay people to write the emulator code.

 

Not that I'm arguing with either of you. Clearly you've both been farther down this path than I have, and your statements are fact, where mine aren't even opinions (maybe more like wishful thinking, evidently).

 

I wasn't aware there was an argument going on. I thought this was a discussion on what we can/cannot do for development of future Flashback systems.

 

Anyhow, if you *can't* do emulation, I think that is a major obstacle to the FB3 that would require you to stick to original hardware. In my opinion, the only way for this to be -largely- commercially viable is for it to execute *original* code.

 

Can you clarify this for me a bit, because it seems like that statement is in conflict. If we use the original game code, and we use a modern design based off of the original system, how is that an obstacle?

[jbanes]

Admittedly, I may be looking at the Jakks product line from a different point of view than the average consumer. I know in the case of the Atari 10-In-1 joystick they did (and assuming on the other games), the games were ported. I do not want to sound like I am against porting. I think it has its place if it is done well.

FWIW, Digital Eclipse does an incredible job with their ports. I have their Joust/Defender for the GBC as well as the Atari 2 Player Paddle games. Both are spot-on recreations of the original games, right down to the HMove lines. So if you do go the port direction, PLEASE hire Digital Eclipse!

[Flojomojo]

Anyhow, if you *can't* do emulation, I think that is a major obstacle to the FB3 that would require you to stick to original hardware. In my opinion, the only way for this to be -largely- commercially viable is for it to execute *original* code.

 

Can you clarify this for me a bit, because it seems like that statement is in conflict. If we use the original game code, and we use a modern design based off of the original system, how is that an obstacle?

Word salad.

 

You can run the original code on vintage, original hardware

You can run the original code on modern, simulated hardware

You can run the original code in emulation, on any hardware

 

Paranoid wants a USB slot which can take ROMs. It really doesn't matter what's running them, so long as it's reasonably accurate. I want the same thing.

[Paranoid]

Yup!

 

*clap clap clap*

 

USB... SD, CF... Sony Memory Stik. I don't care what...

 

If I can buy compilations from Atari and/or download images in ROM format and get them into the machine and have them run, more often than not, and better than worse... I'll be happy.

 

And... Atari's management should pay *careful* attention to this part.

 

By allowing me to put Roms from "the wild" on the machine, they do not *hurt* their potential sales of their own ROM compilations much at all.

 

Instead, they allow me to source Rom images that they do not officially own the licenses for, which is a HUGE value-add selling point for their hardware retro console. Games that otherwise will NEVER likely be available again, remain available.

 

Does that make sense?

Edited December 29, 2006 by Paranoid

[Retro Rogue]

FWIW, Digital Eclipse does an incredible job with their ports. I have their Joust/Defender for the GBC as well as the Atari 2 Player Paddle games. Both are spot-on recreations of the original games, right down to the HMove lines. So if you do go the port direction, PLEASE hire Digital Eclipse!

 

Unfortunately, such power comes with a price (pun intended). Jeff and the guys at DE have been the mainstay for retro collections/ports for a lot of years now and blazed a lot of ground. As you noted, they deserve their reputation. However, companies are looking for other resources as they seek less aggressive contracts, costs, etc. I.E. DE's costs have become as large as their reputation.

 

All I can say in regards to any of this is that you can be sure Legacy will use good, qualified, and passionate people to do the job right no matter what route any of these products goes.

[jbanes]

All I can say in regards to any of this is that you can be sure Legacy will use good, qualified, and passionate people to do the job right no matter what route any of these products goes.

Indeed. Legacy Engineering has been to hardware as Digital Eclipse has been to software. I just hope that Atari recognizes it.

[mstulir]

Word salad.

 

You can run the original code on vintage, original hardware

You can run the original code on modern, simulated hardware

You can run the original code in emulation, on any hardware

 

Paranoid wants a USB slot which can take ROMs. It really doesn't matter what's running them, so long as it's reasonably accurate. I want the same thing.

 

Gotcha. Ok. Sorry for missing the meaning there. It makes sense now.

[ijor]

There is NO SUCH THING as 100% accurate emulation.

 

That depends on the exact meaning of that sentence. Accurate emulation has two (almost) completely different senses. One is the human POV, if he perceives things exactly as in the original hardware or not. The other is the code perspective, if it behaves identically or not.

 

From the human POV, there is definitely no chance to get even close to 100%. But from the technical and code POV, I see no reason why 100% accuracy wouldn't be possible. It is true that we are not yet there, but we'll eventually will. Ok, may be not 100%, but 99.99%. Even with hardware you don't reach 100%. It will probably take a while for 16-bit consoles and computers. But 8-bit ones shouldn't take too much.

 

MAME is very different than single emulation. In MAME there is no much motivation to improve accuracy past certain point. It is much more interesting to add new stuff.