Atmel AVR UC3

[hpglow]

http://www.atmel.com/dyn/products/devices.asp?category_id=163&family_id=607&subfamily_id=2138
Never noticed these before.

Here is an eval kit at Mouser.
http://www.mouser.com/Search/ProductDetail.aspx?qs=bC%252bA5GyyYShiFIv9M353aQ%3d%3d

There is another eval kit that is actually in the shape of a game controler with capacitive touch buttons but it is far more expensive. Could these chips be a beter alternative to the XMegas? Some have built in sound decoders.

[hpglow]

Just got done reading through some of the documentation for this thing. Looks interesting. It can run programs from external memory, and Atmel has a write up on how to do it. I may have to order one of the eval boards and see if I can get the Uzebox ported to it. Although with my history of actually getting projects finished and working this may not be a great idea.

[uze6666]

I think the 32-bit AVR were discussed a long time ago as a potential successor. Now some have up to 128K of RAM which is quite interesting. I know you can use AVR Studio for dev and I'd have to look if it can be programmed with a regular AVRISP device. Frequency is "just" 66Mhz though.

-Uze

[hpglow]

I see your point Uze, but these have 1.5 x the performance per clock than the original ATMegas did. And they have their own audio decoder so that wold be offloaded as well. Still not leaps and bounds, but bigger ram space and external ram options are the main bonus I see to these. It may even be possible to get these to process NTSC signals in software thus removing that expensive chip from the design that alone has to be worth something. As well as the fact that with minor tweeks the current codebase could be compatible with it.

The cons are that it still isn't a major step forward, and they arn't in a package that would make them easy for a hobbiest build. But I think finding a modern microcontroller in pdip packaging is going to prove very hard as time goes on.

I will keep researching, there has to be a good suitor for upgrade.

[DaveyPocket]

It may even be possible to get these to process NTSC signals in software thus removing that expensive chip from the design that alone has to be worth something.

Though if I were to upgrade the Uzebox JAMMA it must have RGB output. SCART users would also be knocked out. I, personally, never liked "internal" NTSC generation on hardware designs, mainly because I can't easily hack my NES or other systems to run RGB output but it's not that necessary.

thus removing that expensive chip from the design that alone has to be worth something.

Most of the time you only buy hardware once, just how I feel. At the same time I wish the AD725/723 chips weren't so expensive for what they are, but it's worth the extra $10.

[hpglow]

I think an ideal situation is one where RGB or NTSC could be processed and the user would be able to choose thier configuration. We of course are still talking legacy video connections, my ideal setup would be to have the video and audio carried over a single HDMI cable over to the TV set. However, due to simplicitys sake it is still advantagous to utilize RGB, composite, s-video, or SCART.

Now that mouser has the STM32 F4 boards in stock for $19. I'm thinking that will be a better option than the AVR UC3 anyway. 168mhz 1mb flash, and 192k of ram, and cheaper per chip than Atmel's stuff. Right now its all about just dreaming of higher resolutions, more colors, and more and larger on screen sprites. After I finish a couple projects I will probibly order one, unless something better pops up. After that I'm going to try and run some feasability tests with it. I just want to see what can be done, if something can be made as a possible sucsesor to the current hardware then pitch it to everyone. Not alienating anyone, and getting compatibility with the current hardware would be first priorities IMO.

Anyway, it is just harmless discussion at this point.

[DaveyPocket]

I'm thinking that will be a better option than the AVR UC3 anyway. 168mhz 1mb flash, and 192k of ram, and cheaper per chip than Atmel's stuff. Right now its all about just dreaming of higher resolutions, more colors, and more and larger on screen sprites

The responses I sometimes get for the Uzebox JAMMA are "specs are too weak". It all depends on what type of game one wants and how good one's programming skills are to conserve storage space and RAM. It would be nice to support better scrolling, more sprites, flipping, resizing, etc... Like you said, people want more power while keeping it simple and "retro".

Anyway, it is just harmless discussion at this point.

Indeed it is.

[hpglow]

I'm thinking that will be a better option than the AVR UC3 anyway. 168mhz 1mb flash, and 192k of ram, and cheaper per chip than Atmel's stuff. Right now its all about just dreaming of higher resolutions, more colors, and more and larger on screen sprites

The responses I sometimes get for the Uzebox JAMMA are "specs are too weak". It all depends on what type of game one wants and how good one's programming skills are to conserve storage space and RAM. It would be nice to support better scrolling, more sprites, flipping, resizing, etc... Like you said, people want more power while keeping it simple and "retro".

Anyway, it is just harmless discussion at this point.

Indeed it is.

It could be a 1Ghz quad core chip for $10 bucks and there would still be some people the complain that it wasn't enough. I would just like to see more people developing software, more activity, that is my thought behind developing something new. I'm not sure if getting a newer stronger chip would do that, but it wouldn't take people away I hope.

[DaveyPocket]

It could be a 1Ghz quad core chip for $10 bucks and there would still be some people the complain that it wasn't enough. I would just like to see more people developing software, more activity, that is my thought behind developing something new.

The reason I chose to work with this project was because of the simplicity: It used AVR Studio, which I already had installed, it was programmed in C, which I already knew, programming was done through AVRISP programmer, which I owned, it was only two chips and a few more components, which I was able to assemble. Like you said with the quad core chip, we could easily make this use FPGA, but that's a little too much for a project like this and not many know how to work with FPGA.

One thing I always thought of if someone owned a Uzebox and wanted to keep upgrading to a minimum cost was to make some sort of pin-compatible attachment to fit the new chip into the ATmega644 40-pin socket, that way one wouldn't need entirely new hardware. Just a thought.

I would just like to see more people developing software, more activity, that is my thought behind developing something new. I'm not sure if getting a newer stronger chip would do that, but it wouldn't take people away I hope.

That may require more advertisement of the project to certain areas like game development forums or other news sites. Possibly a better tutorial base and maybe some better tools for graphics and audio.

[hpglow]

Well the fpga issue is definatly real. My though was that someone would have them mounted to an pdip adapter and they could be sold at the UzeBox store and Sparkfun. Programming through AVR studio is also a big bonus for this project IMO, many of the tools for ARM just arn't that good or they cost money.

The pin compatible option with the current board is real hard. That would leave only two options, the ATMega 1284 pdip package, which is still hard to get and would only be a flash and ram upgrade (perfectly fine with me honestly.) The other being the AVR UC3 fpga (there is one in the series that will take 5.5V) mounted to an pdip adapter. The reason being these are the only two microcontrollers that run on 5V, and most modern controlers take 3.3V. While it is easy to put a regulator on the fpga to pidip adapter board the problem would actually be with the ladder DAC on the main board. All the resistor values would have to be different for 3.3V. I suppose a complete pcb "shield" of some kind could be plugged into the current board but that would up the cost.

I have a couple other ideas for drawing people in. After I get done with Solitare, I'm going to see if I can port miniLZO over to the uzebox. It is supposed to compile to 5k. I'm thinking that if I take out the compression (leaving just the de-compression) from the library it may be smaller. Then just reqire all the files to be pre-compressed on the pc first. Any amount of compression should help make better games, and having a compression library avalible to people couldn't hurt. 5K is a sizeable chunk of flash though, so the compression ratio would have to be worth it. I'm sure the wiki could be better organized, may be a primer for each video mode. I suppose I will look into that before working on a hardware project.