out-of-the-box Petalinux

[wpless]

Hello,

you are advertising the Eclypse Z7 with

'Petalinux is supported o-o-t-b.'

I just powered the Eclipse z7 on and what I got out-of-the-box was just a stupid LED blink and button i was pressed over UART something. 

Can you point me to the document that explains for a newby what steps to take to get Petalinux run on the board?

Thx,

Wilfried

[JColvin]

Hi @wpless,

The Petalinux project for the Eclypse Z7 is available on our GitHub here: https://github.com/Digilent/Eclypse-Z7-OS.

Thanks,
JColvin

[wpless]

and that is what you call "out-of-the-box"?

This starts with a 144 pg document that I have to go through (UG1144), which in turn has some links to follow and some lengthy software installations.

Guess, you guys have a lot of fun, when writing those datasheets :-).

Thanks (for almost nothing),

Wilfried

[Ciprian]

Hi @wpless

Quote

Can you point me to the document that explains for a newby what steps to take to get Petalinux run on the board?

If you just want to run it and make no changes to it you can use Digilents releases for it. Here: https://github.com/Digilent/Eclypse-Z7/releases  v0.2 is the latest and it will contain and .img file which you can flash to your SDcard and have everything set up for you.

Quote

and that is what you call "out-of-the-box"?

Digilent, as far as I know, doesn't provide SDcards with the Eclypse-Z7 "box" but you can have a fully functional debian running on the board, without rebuild, with the above link.

-Ciprian

[wpless]

yes, this was what I was looking for. 

Thanks,

Wilfried

[zygot]

1 hour ago, wpless said:

yes, this was what I was looking for.

I don't know what your expectations for the Eclypse-Z7 are but you will do well to browse through all of the content posted about the board. Just a helpful suggestion.

The sales pitch: "This system allows users to plug in their Zmods of choice and get started prototyping new measurement, instrumentation and control systems without directly interfacing with the FPGA until desired. With this system, embedded Linux developers can leverage the power of FPGA without possessing hardware expertise. Currently, C and C++ are supported, with plans to add support for other programming languages in the future." is still a far off envisioning of the current support. The board is not an Analog Discovery with better capabilities.

Still, it's usable for some applications if you can do your own HDL design work. Expect to do some heavy reading.

[wpless]

Hello zygot,

what I want to do first is using the Eclypse Z7 as replayer of .wav files at highest possible speed. If that is possible, I'd use the Ethernet link to dump the .wav to the Eclypse. If necessary for highest speed, I'd use the SD card.

According to what you were writing, I could probably do that job w/o FPGA programming knowledge, already, right?

This application seems so generic to me that I am assuming that there is something out there, already, which is doing the job. Do you know that?

Thanks,
Wilfried. 

[zygot]

2 hours ago, wpless said:

According to what you were writing, I could probably do that job w/o FPGA programming knowledge, already, right?

Well it depends on what "highest possible speed" is, but yes I don't see a reason to use an FPGA to do what you want to do if we're talking about audio frequency processing... depending on the details. It also depends on what you expect to do with the resulting content. It's not possible to discuss such a loosely defined goal intelligently.

 

2 hours ago, wpless said:

this application seems so generic to me that I am assuming that there is something out there, already, which is doing the job.

I think that this is a bad assumption. I don't think that choosing any specific goal and assuming that you can blindly use an arbitrary flavor of Linux for an arbitrary target architecture and expect there to be readily available code to implement it is a very good plan. Adding a specialized platform like the Eclypse-Z7 into the list of variables just complicates things.

There are a lot of applications that need, or at least can benefit from having, programmable logic available to implement interfaces or accelerate algorithmic processing. That doesn't imply that adding the complexity of FPGA development to your task list is necessarily appropriate. 

[wpless]

Ok, zygot, let me try to explain my application better: 
First: my application is not audio.
With some measurement hardware I am recording up nine channels of analog data. Depending on how many channels I am recording, sample rate can go up to as high as 25 MS/s. The recording is saved in a .wav file. For the development of some specific software of the recording device I need to repetitively replay selected channels of such recordings until this specific software that needs to be devoloped functions to my satisfaction.

Therefore I have to write some software
- that allows a user to select the channels to be replayed (if more than four channels were recorded)
- that decimates the recording to something, the Eclypse can swallow (1 MS/s / ch seems doable)
- that replays the decimated file
 

For me it's probably easier to do the channel selection and the decimation work on a PC.

Preferably, I would use an Ethernet connection to the Eclypse to feed it with the data to be replayed. For maximum replay rate it might be required, though, that I have to store my replay file on an SD card and use that for replaying.

You think that the Eclypse is the wrong platform for such a task? You have a better one (budget might be the limitation)?

[wpless]

zygot, since I intend to do the channel selection and decimation on a PC and use the Eclypse just as a device that feeds the ZMOD DACs at the highest speed possible, I thought that there might be something available more or less off the shelve. Overoptimistic, I understand:-).

[zygot]

1 hour ago, wpless said:

I thought that there might be something available more or less off the shelve. Overoptimistic, I understand:-).

In FPGA development optimism is required; over-optimism is fatal.

But, OK, your elaboration on the project is beginning to make sense. But there are still some details to work out. I've used the Eclypse-Z7 and haven't found it to be the general purpose ZMOD platform that it could be, or that I want. For instance, the vendor support only allows for up to 16383 ADC contiguous samples. Since you are planning on doing processing on a PC I'd suggest looking at the Opal Kelly XEM7320 using Digilent's ZMODs. I've posted about my experiences using this platform here: https://forum.digilentinc.com/topic/20153-capture-4-channels-of-120-million-adc-samples/ and here: https://forum.digilentinc.com/topic/20299-fun-with-phasors/

I suspect that with either FPGA platform you will need to do quite a bit of your own HDL design work. The Eclypse-Z7 ZYNQ just makes this harder and more complicated.

A central issue is having external DDR storage directly available to the logic. For the XEM7320 this is the case. For the the Eclypse-Z7 data has to go to a PS controlled DDR though an AXI bus.

Those are the only inexpensive Xilinx options that I know of ( Opal Kelly does offer its own SYZYGY ADC and DAC pods ). regardless of platform you will have to do the channelization of analog signals on your own. You don't mention what the channel analog bandwidth is, and this would be important information to have for sketching out a preliminary block diagram for a solution to your project.

Basically, I'm telling you to do your homework before starting the platform selection process. There's enough posted about SYZYGY on the Digilent Forums to help get a preview as to what you will be getting into with what's currently available to generate good questions.

 

[wpless]

thanks, zygot. I will do some of the proposed reading and then return back. And even though you are absolutely right with your recommendation to first do the homework before starting the HW selection process, budget situations together with fiscal year ends are sometimes such that you have to make premature decisions.

[zygot]

Nine channels is an unfortunate number. Don't conflate aggregate sampling or data rates with channel bandwidth requirements. There might be better alternatives to trying to sample multiple channels with one converter. Unfortunately, mistakes early on, like system architecture and platform selection, tend to cascade into ever more difficult problems to solve as well as more limited options for resolving them. So, what you do before committing to a path is the most important part of your project.