Zynq devices, boards and suppliers

I'd like to pick people's brains about aspects of
different *suppliers* of Zynq boards. Avnet and Digilent
are front-runners, but any info/opinions about other
suppliers would be helpful too.

   - ease of using their embedded linux. My needs
     are simple, requiring a shell and TCP/IP protocols
     over ethernet. GUI not required, but might be
     used if it didn't complicate the development.

   - quality of online support. How easy is it
     likely to be to find the information so that
     I can (a) duplicate any supplied demo environment
     and (b) mutate it so that my code accesses my
     programmable logic

   - board production longevity. I'm not concerned
     about decades, but I would be concerned if a
     board was unobtainable within months

   - ISE or Vivado environment

Background and context...

I'm intending to develop something based around a small
Xylinx Zynq device. Cost is an issue, but not to the
extent that I will be developing a board containing
the FPGA itself. I will, however, be developing a small
simple add-on board containing my analogue circuits.

Now I can read a datasheet and schematic and outline
to determine the extent to which a board is suitable.
However, as we are all aware, those documents /don't/
cover all the important points when choosing a board!

I've created many stand-alone hardware and software
embedded systems, but *not* based on linux *nor* on ARM
*nor* in the Xilinx ecosystem. Since Zynq devices
represent a complex environment, I'll have a learning
curve (good, I like challenges), and I'm interested
in the quality of the resources and support that
I'll need to overcome my misapprehensions.

Tom Gardner wrote:
> I'd like to pick people's brains about aspects of
> different *suppliers* of Zynq boards. Avnet and Digilent
> are front-runners, but any info/opinions about other
> suppliers would be helpful too.

If you don't need it now, you might take a look at the Parallella board:

http://www.parallella.org

Currently they have some issues with EOL parts and re-designing parts of
the board, so it will be delayed a bit. But I guess there are not much
boards with the Zynq chip for $99, and you get a lot of peripherals,
too, and of course the Epiphany coprocessor. And the Zynq chip used on
the Parallella board is supported by the free Xilinx ISE version:

http://forums.parallella.org/viewtopic.php?f=23&t=134

Linux is already working for the board (Ubuntu, well, you can't have
anything, but should be not too difficult to port Debian for it). Even
if you don't use it for your project later, might be a good starting
point to learn to program the Zynq.

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

On 16/10/13 12:56, Frank Buss wrote:
> Tom Gardner wrote:
>> I'd like to pick people's brains about aspects of
>> different *suppliers* of Zynq boards. Avnet and Digilent
>> are front-runners, but any info/opinions about other
>> suppliers would be helpful too.
>
> If you don't need it now, you might take a look at the Parallella board:
>
> http://www.parallella.org
>
> Currently they have some issues with EOL parts and re-designing parts of
> the board, so it will be delayed a bit. But I guess there are not much
> boards with the Zynq chip for $99, and you get a lot of peripherals,
> too, and of course the Epiphany coprocessor. And the Zynq chip used on
> the Parallella board is supported by the free Xilinx ISE version:
>
> http://forums.parallella.org/viewtopic.php?f=23&t=134

Thanks.

I'd looked at that board a while ago, but at that time it
was insufficiently real. Having seen the standard of some
"open source" hardware[*] I'd like to see it be used in
anger before I committed to it. Contrariwise, longevity might
not be such an issue!

The low cost is remarkable. Before committing, I'd
want to find out:
   - how to disable or ignore the Epiphany processor,
     since it isn't useful to me
   - what proportion of the programmable logic resources
     would be available for me to use for my logic

[*] and some COTS hardware for that matter :(

 
> Xylinx Zynq device. Cost is an issue, but not to the
> extent that I will be developing a board containing
> the FPGA itself. I will, however, be developing a small
> simple add-on board containing my analogue circuits.

This is almost a desktop class chip.  Do you really need that power?  I wish they have something smaller.  I don't need dual core GHz core, but may be you do.

>   - what proportion of the programmable logic resources
>     would be available for me to use for my logic

They says a Z7010 or Z7020.  But for $99, most likely Z7010, which has 28K logic cells available.

If you are not building with chips, does it really matter with:

1. A $50 CPU + tiny FPGA (28K)
or
2. A $30 CPU + $20 FPGA (500K)

I am sure you can find a cheaper board with 2.

On 16/10/13 21:18, edward.ming.lee@gmail.com wrote:
>
>> Xylinx Zynq device. Cost is an issue, but not to the
>> extent that I will be developing a board containing
>> the FPGA itself. I will, however, be developing a small
>> simple add-on board containing my analogue circuits.
>
> This is almost a desktop class chip.  Do you really need that power?  I wish they have something smaller.  I don't need dual core GHz core, but may be you do.

No I don't need that. But if it is there, I may find a use for it.

However, I do need one aspect of the SERDES I/O.


>>    - what proportion of the programmable logic resources
>>      would be available for me to use for my logic
>
> They says a Z7010 or Z7020.  But for $99, most likely Z7010, which has 28K logic cells available.
>
> If you are not building with chips, does it really matter with:
>
> 1. A $50 CPU + tiny FPGA (28K)
> or
> 2. A $30 CPU + $20 FPGA (500K)
>
> I am sure you can find a cheaper board with 2.

So am I. But cost isn't everything; my time is more important.
Hence my question about board suppliers, not FPGA devices.

On 16/10/2013 12:28, Tom Gardner wrote:
> I'd like to pick people's brains about aspects of
> different *suppliers* of Zynq boards. Avnet and Digilent
> are front-runners, but any info/opinions about other
> suppliers would be helpful too.
>
>    - ease of using their embedded linux. My needs
>      are simple, requiring a shell and TCP/IP protocols
>      over ethernet. GUI not required, but might be
>      used if it didn't complicate the development.
>
>    - quality of online support. How easy is it
>      likely to be to find the information so that
>      I can (a) duplicate any supplied demo environment
>      and (b) mutate it so that my code accesses my
>      programmable logic
>
>    - board production longevity. I'm not concerned
>      about decades, but I would be concerned if a
>      board was unobtainable within months
>
>    - ISE or Vivado environment
>
> Background and context...
>
> I'm intending to develop something based around a small
> Xylinx Zynq device. Cost is an issue, but not to the
> extent that I will be developing a board containing
> the FPGA itself. I will, however, be developing a small
> simple add-on board containing my analogue circuits.
>
> Now I can read a datasheet and schematic and outline
> to determine the extent to which a board is suitable.
> However, as we are all aware, those documents /don't/
> cover all the important points when choosing a board!
>
> I've created many stand-alone hardware and software
> embedded systems, but *not* based on linux *nor* on ARM
> *nor* in the Xilinx ecosystem. Since Zynq devices
> represent a complex environment, I'll have a learning
> curve (good, I like challenges), and I'm interested
> in the quality of the resources and support that
> I'll need to overcome my misapprehensions.

You are taking a risk in basing a product on a development board - they 
aren't sold with that in mind and the suppliers don't see long term 
availability as part of the deal. There are quite  a lot of boards based 
on the Zynq around - eg: Mars ZX3- Zynq-7000 SO-DIMM module (which I 
have not used and don't recommend o not recommend.)
You should be able to find a board with long term support and actually 
intended for OEM supply.

Michael Kellett

On 17/10/13 09:31, MK wrote:
> You are taking a risk in basing a product on a development board - they aren't sold with that in mind and the suppliers don't see long term availability as part of the deal. There are quite  a lot of
> boards based on the Zynq around - eg: Mars ZX3- Zynq-7000 SO-DIMM module (which I have not used and don't recommend o not recommend.)
> You should be able to find a board with long term support and actually intended for OEM supply.

Thanks. Your points are well noted. Fortunately it
isn't actually for a product, only a prototype.

My concern is that I want/need decent documentation
and/or community support - and that might not be
there with a board that only has a very short
production life.

Fundamentally I want to grab my data, process it
and display it. I don't want to become an expert
in creating a linux for board Munge v0.9 using
tool Frobnatz v123.456!

Hence my questions about /which/ suppliers/toolsets
that people have found to be "reliable" by
the above definition.

> Thanks. Your points are well noted. Fortunately it
> isn't actually for a product, only a prototype.

We do prototype in hope of building a product someday.  If it's not suitable for a product, it's not suitable for prototype.

> 
> My concern is that I want/need decent documentation
> and/or community support - and that might not be
> there with a board that only has a very short
> production life.

You see community support only if enough people see it as viable tool for projects over the long run. IMHO, the chip is too expensive for most embedded projects.

> 
> 
> 
> Fundamentally I want to grab my data, process it
> and display it.

Sound like all you need are IP cores for SATA, PCIx, etc.

On 17/10/13 15:19, edward.ming.lee@gmail.com wrote:
>
>> Thanks. Your points are well noted. Fortunately it
>> isn't actually for a product, only a prototype.
>
> We do prototype in hope of building a product someday.  If it's not suitable for a product, it's not suitable for prototype.

That might be your position, it isn't mine - and
has never been throughout my working career.


>> My concern is that I want/need decent documentation
>> and/or community support - and that might not be
>> there with a board that only has a very short
>> production life.
>
> You see community support only if enough people see it as viable tool for projects over the long run.

Since that can only be known in retrospect, I'm asking
questions about suppliers.


> IMHO, the chip is too expensive for most embedded projects.

That's a valid judgement, but not the only valid judgement.


>> Fundamentally I want to grab my data, process it
>> and display it.
>
> Sound like all you need are IP cores for SATA, PCIx, etc.

None of those are relevant to me.

Ethernet, TCP/IP and higher level protocol stack are likely
be of interest.

Tom Gardner <spamjunk@blueyonder.co.uk> writes:
> Since Zynq devices represent a complex environment, I'll have a
> learning curve (good, I like challenges), and I'm interested in the
> quality of the resources and support that I'll need to overcome my
> misapprehensions.

I can't suggest any specific boards, but http://blog.elphel.com might be
of some interest.  Its last few entries are about a Zynq-based project.
Maybe posting to the blog's comment section will get some suggestions.