gamebuino mega

[albertinjo]

What do you think of making a version of gamebuino with atmega2560 (in some distant future). 256 KB of flash memory and 8 KB of SRAM sounds nice.

Also is it possible to port the library for the arduino mega board, I was thinking of just remapping the screen SPI pins for the atmega 2560 in the settings part of the library.

[rodot]

I hesitated to use and atmega 644 or even 2560, but they are really really expensive, and I wanted Gamebuino to be as cheap as possible. I might develop a Gamebuino 2 in the future, but I think it would be better to work on expansion modules first as a lot of people request wireless communication, analog joystick, accelerometer, external RAM and sound jack.
But most the things that using an atmega2560 would improve can be added on the existing Gamebuino:
Extra RAM: external 32K SPI RAM module (ok, it's slower)
Extra ROM: use the micro SD card and bootloader
Extra I/O: use a I2C port extender
All these things will be included in the expansion module.
Does that answer your question?

[adekto]

if in a year(or 2) and so want to update the gamebuino
the better chip would be one of the xmega line although that means some architecture changes but the chip is cheaper then a 2560 and can be anything between 4 and 16 kb sram

furthermore it probably be best to update the screen size at that point, since what are you going to do with all that memory

both things will create some problems for backwards compatibility

[Drakker]

if in a year(or 2) and so want to update the gamebuino
the better chip would be one of the xmega line although that means some architecture changes but the chip is cheaper then a 2560 and can be anything between 4 and 16 kb sram

16kb would be amazing,

furthermore it probably be best to update the screen size at that point, since what are you going to do with all that memory

Hmm, stores tons of sprites, smoother animations, 4 levels of parallax, more sounds and music, more enemies and better physics. I can think of plenty of things to do with more memory on a small screen. Right now I feel that we have a very powerful processor that we can hardly exploit to its full potential because of memory limitations, although I might change my mind when I actually get my hands on a Gamebuino. We'll see. ;)

[adekto]

it can do smooth ray casting (3D thing) so 4 layes paralax would be resonable
chip can handle better animations but ur limited by the lcd frame rate (and ghosting)
sound im not sure but i heurd of sd streaming to do digitized sound (also u probely want a dedicated chip if u want beter sound)
physics maybe

ether way i think there is more gain in a biger screen wen ur updating the hardware

[Drakker]

Yes of course we can do it already, but for example, with 4 levels of parallax, you need to feed data for the screen. A full screen is about 500 bytes right now, so if you want 4 levels of parallax (assuming full screen) you need to store a lot of tiles, that fills up the 2kb memory right away, and we don't even have space for actual monsters and hero sprites, the code and variables for physics and etc.

If you up the screen resolution, say, make it 160x144 like a Gameboy, which is about 6 times bigger, you still need to fill it up with sprites, which are going to be larger and again consume all your memory very quickly, and you are left in the exact same situation.

I'd rather have a low resolution color screen instead (if they exist). That would be quite amazing. 2 bits or 4 bits Color sprites use more memory, but you can keep them small enough to keep many in memory. If we want gray right now we need to store sprites in 2 bits anyway, so the memory usage won't be that much higher.

[adekto]

wel ok im not looking at that big of a screen the current coment screen size that is monochrome is 128x64
so its not that drastic increase (baby steps)
ok my problem mostly is on sure u can have parralaxing but whats the point wen u can hardly see it
also u can maybe stream background textures straignt of the sd card
(in fact im looking into how to load sprites ad text in from sd for buinomon project)

[Drakker]

Streaming from the SD card is going to involve some latency, it's good during level load but ideally, once the action start, if you want it to be smooth you will keep SD card access to a minimum... I would not do it for a texture that is drawn every frame.

[rodot]

Communication with the micro SD is pretty fast in fact. Moreover the expansion module will feature 32KB of SPI RAM, which could be used to store bitmaps.
We shouldn't overthink Gamebuino's limitation until we actually tried to develop games on it ^^

And if we really want more power, we can use a cortex M3+, which will provide way more RAM and Mhz for the same price. But it'll break compatibility with arduino. Or we could use the same microcontroller than in chipkit : 80 Mhz 32-bit MIPS 128K Flash, 16K SRAM http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,892,893&Prod=CHIPKIT-UNO32

And if we really, really want power, just develop games on Android hahaha

Edit: If we ever make a Gamebuino 2, we should use an OLED screen to get rid of that ghosting

[Drakker]

I guess we should call on the all mighty Myndale to do some tests for us again, since he has an actual Gamebuino. How long does it take to load, say, 256 or 512 bytes off the SD card, and can the Gamebuino do other processing while waiting for the data, or are we forced to wait for the data once we use the function that loads data from the SD card? If that number is over 25ms, that doesn't leave us much time to compute if we run at 20 fps (50ms per frame), and that's the time normally required for processing a whole frame at 40 fps.

The ghosting is a feature... it's what makes the Gamebuino retro. =)

As for the processor, I think the current chip is fine, it is quite a welcome challenge to code for it. Some more memory would be great, but it is not required to make good games. Although, in the next iteration of the Gamebuino, going for as much memory as possible would be great.