Looking for an extremely cheap FPGA board (in quantity, academic use)

On Sep 10, 2:07=A0am, Simon <goo...@gornall.net> wrote:
> I think you're in luck...
>
> I was browsing around earlier today trying to figure out if I could do re=
altime video-rate JPEG encoding on a DSP instead of having to code it up in=
 verilog, and ran across this...
>
> =A0http://www.nuhorizons.com/development/board.asp?product=3DLattice-ICE4=
0...
>
> For $20, you get:
>
> =A0- 4 capacitive buttons. I guess you still have to 'debounce' capacitiv=
e buttons...
> =A0- 63 i/o on breadboard-friendly 0.1" headers. You'll have to supply th=
e headers.
> =A0- can power off the USB
> =A0- has ~1K logic cells (lut+flip-flop)
> =A0- comes with USB cable & software can be downloaded.
>
> Assuming you can solder some headers onto the board, switches can be repl=
aced by jumpers, and 63 i/o is actually really generous at this sort of lev=
el. Most of the entry-level boards are seriously miserly in comparison.
>
> On the downside, you'd probably not get much of a bulk discount - I'm gue=
ssing they're already cut to the bone at this price...
>
> Just FYI - not a recommendation since I've never used Lattice before, but=
 thought you'd be interested.
>
> Simon

That is _really_ close to what we are looking for, just the headers as
switches thing doesn't quite work.  And I'd really like a bit more
output LEDs. But very encouraging that what we want is possible,
thanks so much for the pointer!

Mark

On 16/09/2012 16:53, Mark Brehob wrote:
> On Sep 10, 2:07 am, Simon <goo...@gornall.net> wrote:
>> I think you're in luck...
>>
>> I was browsing around earlier today trying to figure out if I could do realtime video-rate JPEG encoding on a DSP instead of having to code it up in verilog, and ran across this...
>>
>>   http://www.nuhorizons.com/development/board.asp?product=Lattice-ICE40...
>>
>> For $20, you get:
>>
>>   - 4 capacitive buttons. I guess you still have to 'debounce' capacitive buttons...
>>   - 63 i/o on breadboard-friendly 0.1" headers. You'll have to supply the headers.
>>   - can power off the USB
>>   - has ~1K logic cells (lut+flip-flop)
>>   - comes with USB cable & software can be downloaded.
>>
>> Assuming you can solder some headers onto the board, switches can be replaced by jumpers, and 63 i/o is actually really generous at this sort of level. Most of the entry-level boards are seriously miserly in comparison.
>>
>> On the downside, you'd probably not get much of a bulk discount - I'm guessing they're already cut to the bone at this price...
>>
>> Just FYI - not a recommendation since I've never used Lattice before, but thought you'd be interested.
>>
>> Simon
>
> That is _really_ close to what we are looking for, just the headers as
> switches thing doesn't quite work.  And I'd really like a bit more
> output LEDs. But very encouraging that what we want is possible,
> thanks so much for the pointer!
>
> Mark

Mark - you asked earlier about Lattice Diamond software. I've used 
Lattice parts a lot for the last 5 years and Diamond since it came out.
It works fine, not perfect, and nothing like as many whizzo features as 
the Xilinx stuff but it's cheap and it doesn't crash (at least for me.)

I would much rather work with Lattice parts than X or A, it's all well 
short of the bleeding edge and for just getting the job done at a 
reasonable cost it's fine.

Michael Kellett

Mark Brehob wrote:
>
> Humm, it's sounding like Lattice is going to be the way to go unless I
> can get support from Altera or Xilinx (something I'll pursue if this
> gets out of the "idea" stage).
>
> Has anyone worked with Lattice Diamond?  How does it compare to the
> software from Xilinx and Altera?  I've been burned by some "bad" FPGA
> software in the past (not horrible, just too steep of a learning curve
> for the classroom) and would _really_ like the ability to do schematic
> capture in addition to Verilog...
>
> Thanks again to everyone for their responses, this has been really
> useful!
> Mark
Hello,

Tried to use Lattice software for several years now. It never worked. At 
first it did not work, because my windows installation was too old, and 
they did not have a linux install. I gave it a try at my brother's 
laptop, which was a bit more recent software-wise. Unfortunately their 
license is locked to the boxes' MAC address, so I had to wait for a 
year, to be able to apply for a new one. On that box it did not work 
either... unsure why.
Recent versions of lattice diamond are available for linux, so I gave it 
a try again. Synplify(the synthesis tool) failed with a SIGILL. It turns 
out it requires a SSE2 box.
Synplify is a synthesis tool, that is included in diamond. For some 
parts it is the only practical choice... It is produced by 'synopsys',
and they do not mention any instruction set requirements on their website...
http://www.synopsys.com/Tools/Implementation/FPGAImplementation/FPGASynthesis/Pages/FPGAPlatformSupport.aspx#requirements

The other synthesis tools supported are:
Precision: 3rd party (`mentor graphics`), not included, MachXO2 not 
supported.
Lattice LSE: Lattice Synthesis Engine, included. Available only with 
MachXO, MachXO2, and Platform Manager.

I did not try those, as they don't support LatticeXP parts... they 
_probably_ work.

All in all, things have improved. It might work on your common, everyday 
box.. if it isn't too old.

regards,
JK

On 9/17/2012 12:43 AM, Johann Klammer wrote:
> Mark Brehob wrote:
>>
>> Humm, it's sounding like Lattice is going to be the way to go unless I
>> can get support from Altera or Xilinx (something I'll pursue if this
>> gets out of the "idea" stage).
>>
>> Has anyone worked with Lattice Diamond? How does it compare to the
>> software from Xilinx and Altera? I've been burned by some "bad" FPGA
>> software in the past (not horrible, just too steep of a learning curve
>> for the classroom) and would _really_ like the ability to do schematic
>> capture in addition to Verilog...
>>
>> Thanks again to everyone for their responses, this has been really
>> useful!
>> Mark
> Hello,
>
> Tried to use Lattice software for several years now. It never worked. At
> first it did not work, because my windows installation was too old, and
> they did not have a linux install. I gave it a try at my brother's
> laptop, which was a bit more recent software-wise. Unfortunately their
> license is locked to the boxes' MAC address, so I had to wait for a
> year, to be able to apply for a new one. On that box it did not work
> either... unsure why.
> Recent versions of lattice diamond are available for linux, so I gave it
> a try again. Synplify(the synthesis tool) failed with a SIGILL. It turns
> out it requires a SSE2 box.
> Synplify is a synthesis tool, that is included in diamond. For some
> parts it is the only practical choice... It is produced by 'synopsys',
> and they do not mention any instruction set requirements on their
> website...
> http://www.synopsys.com/Tools/Implementation/FPGAImplementation/FPGASynthesis/Pages/FPGAPlatformSupport.aspx#requirements
>
>
> The other synthesis tools supported are:
> Precision: 3rd party (`mentor graphics`), not included, MachXO2 not
> supported.
> Lattice LSE: Lattice Synthesis Engine, included. Available only with
> MachXO, MachXO2, and Platform Manager.
>
> I did not try those, as they don't support LatticeXP parts... they
> _probably_ work.
>
> All in all, things have improved. It might work on your common, everyday
> box.. if it isn't too old.
>
> regards,
> JK


I have no idea why you could not get their tools to work.  I have never 
had trouble installing the tools other than the license issues which are 
common to all of these vendor supplied tools.  I have also never had 
trouble with getting duplicate licenses for new computers.  I don't know 
why you felt you needed to wait a year.  In fact, when I ask for a 
license renewal, they send me a license file for every machine I have 
ever registered with them!

The only real issue with any of the vendor packages are the IDEs which 
can be different.  Typically they try to emulate a Windows look and feel 
with some sort of a file manager-like GUI and some icons for the various 
files and/or processes of compiling code to produce a bit stream.

The "classic" ispLever is a little clunky in some ways, but it has never 
gotten in the way of doing work.  I have yet to try Diamond.

As to using schematic for any part of a design, I would point out that 
schematic for even top level design is rarely used in industry.  It has 
notable disadvantages with the lack of potability at the top of the 
list.  I suppose I can see why you might prefer it for teaching since it 
is very visual, but in the "real" world this will not be encouraged. 
Structural HDL is not hard to learn or use really.  So why teach 
techniques that are unlikely to be used?

Rick

Mark Brehob wrote:


> Yeah, sadly for us we have no control at all over what computers
> people use, so we can't rely on parallel ports...
OK, one other thing is that due to export restrictions, no vendor's
free software can run on 64-bit operating systems.  There are ways to
hack links to libraries on Linux to get around this for Xilinx web pack
software, apparently.

Jon

On Thursday, 13 September 2012 20:03:07 UTC+1, Jon Elson  wrote:
> rickman wrote:
>=20
>=20
>=20
>=20
>=20
> >=20
>=20
> > If you build the Parllel Cable III into your product, what will you
>=20
> > connect it to?  I don't think they have built a PC with a parallel port
>=20
> > in a number of years and my understanding is that the drivers for USB
>=20
> > parallel ports don't work properly with bit banging software like this.
>=20
> >   Am I mistaken?  Is this a workable solution?
>=20
> I have no idea about the USB-parport, but you CAN get computers with real
>=20
> parallel ports, just not on laptops.  I use the parallel port for lots
>=20
> of interconnect projects, and the Intel D525 chipset still has parport
>=20
> support, if the motherboard maker chooses to bring it out.  For this
>=20
> reason, we use a lot of Intel D525 (Atom) motherboards in projects.
>=20
>=20
>=20
> Jon

I've been burnt by this. Xilinx use a proper plug and play driver for their=
 parallel cable IV and will therefore work with a very cheap PCI parallel p=
ort card. ALTERA don't, their driver insists on the parallel port being whe=
re IBM put it in IO space 30 odd years ago.

Colin

colin <colin_toogood@yahoo.com> wrote:

(snip on FPGA programming)

> I've been burnt by this. Xilinx use a proper plug and play 
> driver for their parallel cable IV and will therefore work 
> with a very cheap PCI parallel port card. ALTERA don't, 
> their driver insists on the parallel port being where IBM 
> put it in IO space 30 odd years ago.

Go through the binary code of the driver and replace the port
number with the desired, different, port number.

I suppose that could also change something else, but it might work.

-- glen