Open Research Institute Weekly Reports, 24 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 20210824 FPGA Standup

Description

FGPA Stand-up meeting for 24 August 2021.

All welcome at Getting Started http://openresearch.institute

A

Good hi, andre hello there, hello, hi, hi, welcome.

B

C

All great, I think, the time to be outside for my garden. If they're gonna show.

A

Never a dull moment right all right. Well, let's go ahead and start uh welcome to the fpga stand-up meeting for a phase four ground from open research institute for tuesday august 24th 2021. um So please let us know what you've done for the past week. What are your plans for the next week? If you have any challenges, if you need anything, any resources or materials andre? Why don't you start us off.

B

um So I did feel I made a little.

B

A wiki page for the petal linux. Basically, the commands are used and things like that, and it's not really meant to be a complete. It's just. You know I did this um and it worked.

B

There is one sort of troubleshooting thing that I sometimes have to reset, and you know just you know things I have to do when using better linux and booting it.

B

I did try building the encoder for the zcu 106 it it worked. I'm trying to figure out um how to use adma because especially reading data, it's very, very slow, so yeah, I'm trying to work out um basically a faster workflow and we're gonna need dna anyways. So um we're still working on that and that's it really.

A

Well, that's plenty yeah, that's weird that it's slow! It's not supposed to be how, when you say slow how slow do you mean.

B

uh So it's it I'm using a axis stream fifo. I think this name of the name of the card and basically, if I want to transmit, I will write, say the length and write to the same address. You know word word, word word, you know to fill in those that that length and then when this completes it will transmit and then on the receiving end is sort of uh mirrored. So I read how many entries there are and then you know, read, read read, but the read frame is way bigger, so reading takes.

B

I don't know a minute.

A

B

Okay, yeah, I I I don't know exactly well the arm is it's not that the arm is slow, but doing this transfer over the arm might not be. You know the fastest thing. It would be way easier if we could dma somewhere and then the arm can just say you know mem up.

A

Right, okay, that makes.

B

A

Yeah, I guess I'm just kind of surprised that it's so slow, but it I mean, since we want to go in a different direction. Maybe we should.

C

A

uh Help you get there yeah.

B

Yeah, um I'm still getting started with dma. I need to first choose one of the four. I think four offerings.

A

B

Yeah that xynx xynx have and yeah then build and start testing.

A

Okay, yeah, let me know if I can help out in any way um and then.

D

A

I'm trying to use the zcu 106 to learn how to how to better learn how to how to use pedal linux and what I've done is I've changed it to where it boots from the sd card. If I need to change it back every time to jtag, which is where it was, let me know so I can. I can do that like when I leave it. I can change it back to to the jtag. I think that it might work. If I have it set to sd card, it should still work from uh jtag.

A

That was something that that I think we discussed for the other board, um but yeah.

B

It I I yeah, I was thinking, I was reading the slack messages and I think it there's no reason why it shouldn't work.

A

B

A

It doesn't just let me know, and then.

C

A

I'm only going to show up and use it when I have to walk through um the next step of both either the training or trying to get.

D

A

To speed with uh with using pedal linux and and using um the sd card approach, so it's easy for me to put it back and leave it there and since most of the time, I think I'm going to be working on it when, when you're on hours, that you would definitely not be.

B

A

So I think this should work out, even if I have to change the um the switches the settings then no problem.

B

A

Just let me know if you have any trouble whatsoever and then I can just leave it for jtag uh every time.

B

Yeah- and there is a way to sort of out like a lock mechanism, because I think you you cannot have- cannot have two people using the same tty usb thingy.

A

Yeah it'll show up. We found this out um yeah.

D

A

Show up as busy or locked it'll say you know: it'll wave you off, so we we did see that yesterday.

B

Yeah, so that's sort of the lock. You know someone is using.

A

Yeah we we did find out that you can't use minicom. You have to use something like screen, so that was that was an interesting lesson. Yeah yeah.

B

I used, I was lucky. I used screening.

A

Yeah, you did good, I think. Oh, my goodness, I we spent quite a bit of time trying to figure out why we couldn't get a serial port, so.

D

A

uh Good we're in good shape now so it's uh you know, moving forward.

B

Nice, nice cool.

A

All right, do you need, do you need any stuff, any things any help or or any particular roadblocks that can help with.

B

A

Yeah I hear you: I've been working really hard on trying to figure out how to how to get a shipment of time in for everybody, so I'll keep at it. Yeah.

D

A

Yeah, I will I'll go ahead and say that over the past week um uh it turns out that uh having my middle daughter go off to college took a lot more time than I thought so, I'm behind where I wanted to be from last week, however, there has been a lot of progress, so the what I did is um started a for real trying to learn how to use pedal linux and trying to get the tools uh under my fingers and and that's coming along.

A

um So that's been good, but I have not yet gotten the other board the zc706 to produce any um signals over the error yet so that was one of the things I wanted to do over the past week and this still on the agenda.

A

um Some some places that I didn't think I would make progress over and I did are working with the m17 team to get them more aware of fpga resources and getting them more aware of hdl and doing more work ahead of time like with bit error rate schemes bit error rate testing to have their kind of ducks in a row, and that's come along very well.

A

Over the past week, so I was expecting much more of a battle there, but there's several people on their team that are that are very interested in moving to fpga and getting more of their work um in the remote labs and on hdl. So that was a surprisingly quick bit of progress that I can report that I expected to take many more weeks.

A

The then the the getting the additional resources online, not getting communications with florida, not yet where they need to be so we're still talking about moving and the the new lab location is not going to be online until september anyway, uh so I'm still trying to get things to additional resources online. In the meantime, that's not progressed over the past week as quickly as I uh thought it might, um but understandable so still, working on that and yeah time uh yeah.

A

The biggest uh challenge for me is logistics and getting more hours out of the day, so a conundrum that we all share. Okay, so that's it for me so off to paul kb5mu. You have the floor.

C

Hello from the remote labs, I have nothing really to report so uh we're trying to provide whatever support is necessary.

A

It's all good right. That sounds sounds like a ringing endorsement of goodness.

C

No news may be good news unless you're, just not paying attention.

A

Well, we okay! So what about the dhcp edition? That's coming.

C

Well, if necessary, I was resisting having a dhcp server on the on the lan, because having static, ips really helps with the automation, but if the pedal linux build can only work with dhcp, then we can. We can compromise on that and add a dhcp server I'd rather figure out how to assign a static ip to the petalinics, build frankly,.

A

Okay, yeah, that's probably that's my fault, because the what I the reason that this happened is because I was using the provided image from the dev development board and it goes out and looks for dhcp so and I think andre spoke up too so so I'm gonna uh defer to him.

A

B

Yeah, that's a good like I kind of forgot about this, because I you know I didn't look into this for a week, but you can set up the static ip in the pedal union's config. So you, basically, if you boot, whatever it will sort of always have that ap and which is yeah very, very nice. I think okay, is that built into.

C

The pedal sorry.

A

Oh no, I was just asking if that's the preferred way forward, because then then I you know it seems like something I should be able to do in building the next image that I want to use so that it seems like that's the right. The right answer for us in the in the remote lab.

B

Yeah, I just wait for the chat I just use it. I I paste the link. I literally did this the thing they're describing so paralleling config and then inside subsystem ethernet obtain ipm. I address automatically. No, so I'm looking at the page. Maybe not everyone is looking, but basically there's. You know the steps and you set up the ap, the mask and the gateway, and that.

A

Works, okay, got it yeah. I got the link. Thank you.

C

Is there a way to uh to set the ip on a pre-built version, or do you have to go all the way back to the pedal linux and rebuild.

B

So yeah, if you have an existing project, you can run the pedolinus config change this and then pedal linux build and it will be fast ish. I think it will be like a minute or so.

B

A

B

A lot of stuff.

A

Yeah, oh yeah,.

C

I don't know anything about linux, config.

A

Yeah, no, it looks like there's some. Oh sorry, paul I didn't mean to talk over you.

C

Does the pet linux config take its starting values from the existing image, or do you have to feed it a file that replaces all the settings it's? What I'm interested in is not so much performance. I want mistakes, but how certain you are you got everything right and the virtue of using a pre-build, pre-built image is that you know everything is was right at some point.

B

So it will generate files, um it's like the so last time I built kernel was I don't know many years ago, but you know how it generates like a config file with everything you changed, so it will generate that and you could in theory, just commit that somewhere, uh meaning part. You know it can be done.

B

You know more automatically than just using the gui to change stuff.

C

C

I don't know enough about the situation to ask follow-up questions, so let me learn a little bit more about pedal linux and maybe we'll address this again.

B

Yeah yeah, I think, do we need a sort of a workflow will sort of emerge.

A

Yeah, no, it would probably help I mean I can see some things in the page that you linked, which is super helpful, and the the first workaround um that's mentioned is something that we actually did yesterday in order to just fix it, which was ifconfig.

C

A

Know like just assign it when you, when you bring it up and use it yeah, you know okay. So if we want to automate that, though it looks like yeah, we have to just rebuild rebuild the package, and you know what that's okay too, like we, we should be able to do this, um so I think we're in good shape. I think we are we'll move towards making it easier to use and easier to to handle, um and then this page is super helpful.

B

And I think the petal linux, the say the operating system as a whole, it's more or less independent of the fpga thing like it's a car, a board thing. uh I I think I might be talking.

B

No, maybe not like it has to know the addresses.

A

B

A

Yeah yeah, I think it does have to yeah. I think we got. We got a good path forward, we'll be able to figure it out.

B

Yeah yeah yeah I'll try to work out how to sort of automate this. So we can, I don't dot dot, slash, run, dot, h and it will just build the thing um you know. So people don't need to go into the config and you know manually change it. It's that's error prone and all.

A

Right yeah, it kind of, is I mean it's, it's okay for us to kind of like clap things together and get things to work.

D

A

You know if we're, if we're all up in the you know all all up in the guts of it, um yeah I'll I'll work with paul and try to figure it out and uh and just take this. This conversation as a good uh basis for, for you know, future work over the next week or so.

B

Cool yeah cool.

A

All right any other comments.

B

If, if there's anything, I can do in the m17 things like you know, just let me know.

A

Okay, I sure will yeah I'll uh absolutely. Will I'll I'll count you in.

C

A

Nice no happy to help all right good deal, see you on slack and.

D

A

Yeah and then we should be able to meet next week and I'm looking forward to it.

B

A

You bet all right see you soon.

B

See ya, thank you.

D