Reconfigurable Computer

[shs2017]

Hello Electronic Forums People,
I'm developing a reconfigurable computer using a fpga but there is one problem; I don't know what fpga to use. Even though it might not be a very popular (but who knows) project, I hope that I can build it to be usable. My definition of usable is:

1. Is fast (either by parallel processing or speed)
2. Is small in size
3. Is reliable

Thank you,
shs2017

 

[davenn]

hi there
welcome to the forums

define what your definition of small in size is
hand held say up to I-pad size ? can be bigger ?

what functions does it need to be able to accomplish ?
What I/O features does it need ?

Dave

 

[shs2017]

I would like the fpga to fit on a phone size but can be bigger. I'm building a reconfigurable computer so it'll need an fpga but one thing I want is not any dedicated hard ip because I want it to be flexible and the hard ip takes up space that could be used for other i/o. It needs to be either very fast or have many pins.

By the way are you Dave from EEVBlog?

 

[BobK]

What do you mean by a reconfigurable computer? The essense of a computer is that it is reconfigurable via programming. Are you talking about something different than this? Do you expect to re-program the FPGA on the fly? Why? What is it you are really trying to do?

Bob

 

[shs2017]

http://en.wikipedia.org/wiki/Reconfigurable_computing

Does anybody know an fpga that would fit the description described in the earlier posts?

 

[davenn]

http://en.wikipedia.org/wiki/Reconfigurable_computing

Does anybody know an fpga that would fit the description described in the earlier posts?

well my initial thoughts are that this would not be doable by the home hobbiest as it would require specialist PCB design to accomodate hi density SMD components along with the several Large Scale Intergration chips to get it into the sort of case dimensions that you are invisioning

cheers
Dave

 

[BobK]

Your best bet would be a development board from Xilinx or Altera. They can be had for < $100.

Bob

 

[shs2017]

http://www.copacobana.org/. Well this is pretty simple and somebody made the boards. I think I can make a better design though but still it's not that hard to produce. If anything I'd be worried about the price.

 

[rob_croxford]

This is a huge project.

Having undertaken a project at uni which involved developing a CPU (in essence) as a year long group prject i speak from experiance. You will need a decent software package (to generate decent simulations) we used Altera for all of our VHDL development and simulation. I think this software is quite expensive but im sure there are cheaper alternatives.

Your best bet would be a development board as bob said but even then getting the thing actually up and running will be a long (none the less rewarding!) task.

A little word of warning we spent 12 months of development in a group of four id say arround 15 hours a week. At the end of this time we had a fairly functional CPU running in software simulations but got no where near developing a PCB or even programming onto an FPGA.

Things you will need:
1) - High level of digital logic understanding
2) - high amounts of patience.
3) - i would stick with VHDL and use a software package that allows you to develop blocks of code and generate physical logic blocks. (for instance you can develop the code of an Adder put this into a block and then simulate it by "physically" connecting those blocks together) allows you to veiw what is happening alot easier than staring at code.
4) - before getting started use google to find out absolutely everything you need to know about what you are trying to design. These sort of projects can lead you down the wrong road for weeks at a time.

All the best!!

Rob

 

[CocaCola]

http://www.copacobana.org/. Well this is pretty simple and somebody made the boards. I think I can make a better design though but still it's not that hard to produce. If anything I'd be worried about the price. .

Pretty simple? Do you have any experience in PC board design and layout?

That example the FPGA boards you linked are easily many 100s almost certainly 1000s of hours worth of layout to get it all that compact and clean, FYI it's 4 layers so you only see half the layout in the pics... Then you have the actual controller board, that is even more complicated and again likely 1000s of hours worth of work...

Hardly my definition of pretty simple

 

[shs2017]

Honestly I don't think it's thousands of hours of work maybe a couple hundred but I'm wiling to spend time developing this board. I've come up with a solution to lower costs, make the board more efficient, and to simplify the design. The idea is to start out with a prom (load bit files), clock (to send and receive signals), power, test led, and the fpga (probably a spartan 6) on one board. Then, later I can add on boards for other features like graphics, audio, memory, and more. Although it would be nice to make a little money on this project it is mainly for myself so time is not an issue. I have experience with developing a operating system, with circuitry, programming (about 15 languages including x86 and arm assembly), and I hope to turn all of these into one project and making it my own. I know many of you disagree with my doing this saying it's to long, takes to much time, or what ever but I think that it would be a good learning experience for myself. I'm pretty set on developing with a spartan 6 but if you have any tips on hardware development with fpga's and which one would be the best choice for the things I described above post it/send me the link/etc.

Thanks

 

[CocaCola]

Honestly I don't think it's thousands of hours of work maybe a couple hundred

Do you have any PC board layout experience, or are you just taking a wild guess based on what you imagine? A highly skilled board layout person that is laying out boards everyday 'might' be able to do it in a few hundred hours, if he was provided a proven good schematic to work from... An inexperienced person or a person that has experience but doesn't do it all the time will easily gobble up exponentially more... I have spent 100s of hours on boards significantly less complex only to have it produced and have to spend more time because of an error or noise issue with the board layout or whatever... CAD software makes in infinitely easier than drawing it on acetate with markers but it's still not a cake walk when the density of the board is this high, and once you get past 2 layers of a high density layout it gets even more complicated...

This thread is going on two weeks now, how much have you accomplished? How many hours have you 'thought' about this project during those two weeks? I'm willing to bet that if you are actually devoted you already have at least 100 hours into this and you are still at square one... The PC board design will be the same way...

 

[shs2017]

I haven't started the design yet because of school. I was just asking for opinions. BTW have you read my post above. The design will be very simple.

 

[CocaCola]

BTW have you read my post above. The design will be very simple.

Yes, I have read your post that is what I base my replies upon... You can say it's simple but it's not, high density small boards are NOT simple, quite the contrary...