recomendation on a processor core

hi,

after few years i want to restart FPGA development once again - initially f= or personel fun, but later may move in to commercial development too. At th= e moment i am trying to find a suitable processor core for the work. These = are my requirements:

  1. It must have a good documentation + tools + community support
  2. It should be mature and future proof
  3. Better if it is Free, how ever commercial cores are OK too provided that= i have the full documentation + tool set for free to get started.

Few years back i played a little with openrisc. This time too i initially w= anted to go with it. However documentation for openrisc seems to be extreme= ly low and incomplete. Im not sure the status of tools and support, but see= ms like they are lacking as well. I'd like to get some advice from expert e= ngineers about this before going forward. Also like to know about better al= ternatives as well.

thank you

Reply to
manusha1980
Loading thread data ...

personel fun, but later may move in to commercial development too. At the moment i am trying to find a suitable processor core for the work. These are my requirements:

have the full documentation + tool set for free to get started.

wanted to go with it. However documentation for openrisc seems to be extremely low and incomplete. Im not sure the status of tools and support, but seems like they are lacking as well. I'd like to get some advice from expert engineers about this before going forward. Also like to know about better alternatives as well.

How fast do you need it? I've used NIOS, the smallest version is free. Sometimes a bit difficult to setup a project and to use Eclipse, but when it works, it is nice, with integrated debugger.

--
Frank Buss, http://www.frank-buss.de
electronics and more: http://www.youtube.com/user/frankbuss
Reply to
Frank Buss

ly for personel fun, but later may move in to commercial development too. A= t the moment i am trying to find a suitable processor core for the work. Th= ese are my requirements:

that i have the full documentation + tool set for free to get started.

ly wanted to go with it. However documentation for openrisc seems to be ext= remely low and incomplete. Im not sure the status of tools and support, but= seems like they are lacking as well. I'd like to get some advice from expe= rt engineers about this before going forward. Also like to know about bette= r alternatives as well.

ly for personel fun, but later may move in to commercial development too. A= t the moment i am trying to find a suitable processor core for the work. Th= ese are my requirements:

that i have the full documentation + tool set for free to get started.

ly wanted to go with it. However documentation for openrisc seems to be ext= remely low and incomplete. Im not sure the status of tools and support, but= seems like they are lacking as well. I'd like to get some advice from expe= rt engineers about this before going forward. Also like to know about bette= r alternatives as well.

i'd like to have something that i can use for serious stuff, for example mp=

3 encoding/decoding, implementing neural networks, decoding movies. I start= ed out with leon2/leon3 and things have been good so far. but still interes= ted in knowing alternatives.
Reply to
manusha1980

If you have a low end FPGA like Cyclone, I think just a CPU imlemented in the FPGA will be too slow for things like decoding movies, but you could add FPGA logic for some tasks. With NIOS you can embed it as custom instructions, but the usual way is memory mapped. Of course, the easiest solution would be to buy a Raspberry Pi for $25 for the tasks you've mentioned, except the neural networks, which could be faster with FPGAs, because of the high parallel processing capabilities and the simple implementation of one neuron, but take a look at GPUs too for it.

--
Frank Buss, http://www.frank-buss.de
electronics and more: http://www.youtube.com/user/frankbuss
Reply to
Frank Buss

If you're using a processor core for serious stuff, instead of using it to control the serious stuff, you're probably missing the point of an FPGA...

- Brian

Reply to
Brian Drummond

You can use a NIOS for serious stuff - you are not going to get anything significantly faster in an FPGA soft core (I expect Xilinx's soft core is similar). MP3 encoding/decoding is not /that/ demanding. Neural networks, of course, scale from something that can run on a tiny microcontroller to networks that take forever to run on the biggest supercomputer, so they are a useless example.

Decoding video is not going to run in software on a 150/200 MHz processor - you do the hard work in the FPGA itself. The trick here is to bind it well with the NIOS - perhaps using custom instructions, or a memory-mapped peripheral. You can even use Altera's tools for turning C code into FPGA hardware.

There are also some FPGA's with hard cores (PPC or ARM). These will run a lot faster, but cost a lot more.

Reply to
David Brown

The II/s and II/f maybe. But the free II/e needs at least 6 cycles per instruction. Altera's benchmarks on it are in the 20-30 MIPS range for Cylone III, and even on a Stratix V don't get up to even 50 MIPS. Sure they're tiny, you can parallel a mess of them, but they're really meant for occasional control plane stuff rather than heavy lifting.

--
Rob Gaddi, Highland Technology -- www.highlandtechnology.com
Email address domain is currently out of order.  See above to fix.
Reply to
Rob Gaddi

Well, yes, I was referring to the "real" Nios's - the ones designed for fast, efficient processing in a soft core, rather than the version aimed at minimal space. As you say, the II/e has its usage - but not as a fast processor.

Reply to
David Brown

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.