<charlottia>
<attiegrande[m]> "whitequark (@_discord_1821742088..." <- confirming receipt on the Matrix side! I believe the reason you aren't seeing Cat in the Discord channel member lists is because it uses the `irc-only` role to explicitly not be in the amaranth/glasgow channels on Discord.
<attiegrande[m]>
thanks @charlotte - we figured out that the puppets leave on extended periods of inactivity, which is fair enough
<charlottia>
Right, I see now! I didn't actually click over to read the backscroll in #glasgow-interface-explorer:matrix.org yet >_>
cr1901 has quit [Read error: Connection reset by peer]
cr1901 has joined #amaranth-lang
GenTooMan has quit [Ping timeout: 240 seconds]
GenTooMan has joined #amaranth-lang
Degi_ has joined #amaranth-lang
Degi has quit [Ping timeout: 246 seconds]
Degi_ is now known as Degi
jevinskie[m] has joined #amaranth-lang
<jevinskie[m]>
I’d hate to see LUNA/Cynthion fade away. I had some great success using LUNA with my DECA board and integrating it with LiteX :( https://github.com/jevinskie/liteluna
<jevinskie[m]>
I did have to buy a beagle 480 to debug my LUNA bringup but not due to any issues in LUNA itself but my LiteX gateware wrapped around it :P
indy_ has joined #amaranth-lang
nak has joined #amaranth-lang
nak has quit [Ping timeout: 240 seconds]
mindw0rk has quit [Ping timeout: 245 seconds]
mindw0rk has joined #amaranth-lang
mindw0rk has quit [Read error: Connection reset by peer]
<anuejn>
speaking of LUNA: did anyone of you have sucess using USB-3 with the ECP5 transcievers?
<anuejn>
also: does one need to support USB-2 HS for being able to negotiate SuperSpeed? (so is it theoretically possible to leave out the USB-2.0 PHY)
miek__[m] has joined #amaranth-lang
<miek__[m]>
no, you don't need to support USB2 HS for SuperSpeed, the buses are completely independent (and yeah, you can leave out the USB2 PHY)
<miek__[m]>
USB3 on the ECP5 is still experimental, in my testing it can enumerate but doesn't stay connected for long after that
miek[m] has joined #amaranth-lang
<miek[m]>
<jevinskie[m]> "I’d hate to see LUNA/Cynthion..." <- Don't worry, they're not going to fade away. I'm not really sure where all this is coming from, we've got several people in GSG quite capable of developing & maintaining them, and we'll be doing that for the foreseeable future
<Wanda[cis]>
mcc111: `do_program=False` will still generate the bitstream; what you want is `do_build=False`
<Wanda[cis]>
though... um
<Wanda[cis]>
I'm a little hazy on the details of how to use it
<tpw_rules>
anuejn: yeah i have devices that don't bother wiring up the usb2 wires and work fine
adamgreig[m] has joined #amaranth-lang
<adamgreig[m]>
mcc111: if you're building for ice40, the simplest thing is probably to just let amaranth produce the bitstream for you, rather than going through your docker container business, if it's an option?
<adamgreig[m]>
if it's an ice40up5k then amaranth can certainly target the chip, and targetting the board just needs you to write a platform class to tell it which pins have which names, which you'd need to do anyway to use platform.request
<adamgreig[m]>
I'm not completely sure but I don't think it makes sense to be combining verilog.convert() with platform.request(); if you're converting to verilog you should just be exposing all the IO names as ports on the top-level object which your toolchain will eventually connect to the right pins using whatever constraint file it has
<adamgreig[m]>
(so, verilog.convert(ports=[top.led, top.switch, ...] sort of thing)
<adamgreig[m]>
but yea, I'd defo just skip the docker business and use amaranth's build system directly, it should be much easier
<anuejn>
can you share which devices / do they work somewhat stable?
<tpw_rules>
anuejn: flir boson thermal cameras. they are totally stable, aside from power related issues. completely unrelated to LUNA/ecp5, just saying that it works to not have usb2 wired up at allo
<anuejn>
ah I see
<tpw_rules>
idk if it's Standard Compliant, these are research instruments and a bit hacky themselves
<tpw_rules>
but i haven't bumped into problems with that particular aspect on various versions of windows and linux
<tpw_rules>
(they are also fine running usb3 over 1mm jst connectors)
<anuejn>
nice one
<anuejn>
but do they also not implement usb 1?
<tpw_rules>
correct, the usb 1/2 D+ and D- pins are simply not wired to the connector
<anuejn>
thats a bold move
<tpw_rules>
yeah, it's a little hard to believe. they do have usb2 available on a separate connector so the firmware supports both. but they go to different usb plugs and i imagine weird things would happen if you plugged both in at the same time
<urja>
i mean, afaik it is also possible (though, depending on host i suppose) to plug two devices into an usb3 port (one on the usb1/2 port, one on superspeed...)
<mcc111[m]>
<adamgreig[m]> "you basically just need https..." <- Thanks, I may ask for help or at least ask you to review my attempt.
<mcc111[m]>
The reason I'm nervous about bringing my own compiler is the board has their own vended (forked?) version of icestorm . I don't have the background to know how much difference that could possibly make.
<adamgreig[m]>
It seems like it could only be worse
<whitequark[cis]>
the board should not be doing things like that
<whitequark[cis]>
there is literally 0 reason for it to have a fork of icestorm
<whitequark[cis]>
it does not matter what board it is, this is simply not how anything works
<adamgreig[m]>
Yea, it would be like if they provided a gcc fork to use with their off the shelf arm microcontroller
<adamgreig[m]>
I suspect they've just tried to make a convenient docker based package
<whitequark[cis]>
anuejn: yes @VioletEternity brought up USB on ECP5 transceivers and it works perfectly fine, I've seen it transfer 10 GB in a row with no errors
<adamgreig[m]>
Rather than guiding all their users through installing icestorm years ago which was a little involved
<whitequark[cis]>
what miek says was the case before her patches (which should now be upstream in LUNA)
<adamgreig[m]>
But since you can get binary builds of icestorm+nextpnr, or even use the yowasp python/wasm versions, and then it actually integrates into amaranth, you'd just be making a lot of extra work to use the packaged (and outdated?) docker
<whitequark[cis]>
yeah if you're on a mac machine, just use yowasp?
<anuejn>
whitequark[cis]: that sounds extremely cool
<mcc111[m]>
<whitequark[cis]> "the board should not be doing..." <- OK, good to know, thanks. I only got worried because this board appears to be doing at least one other thing it shouldn't be doing
<whitequark[cis]>
which thing is that?
<miek[m]>
fwiw I have been using those patches, but I haven't been able to reproduce that level of reliability so far
<whitequark[cis]>
interesting
<whitequark[cis]>
I know she got it to be that reliable on all of the boards she had, with three different hub ICs and on, I think, two HCs
<whitequark[cis]>
I wonder what's up with that
<mcc111[m]>
<whitequark[cis]> "which thing is that?" <- Well, what I'm thinking of is they somehow broke some of the macros in the Arduino SDK. I don't know if they actually changed something or if they just configured it wrong, like set up Arduino to target one chip and they're using another.
<mcc111[m]>
Anyway, I'd definitely prefer to get onto the Amaranth build system as soon as I can, so I'm glad to hear it's expected to work.
<whitequark[cis]>
ah
<jevinskie[m]>
BTW mcc11 gtkwave this month did a ton of refactoring so that may be related to your issues. I can try to get a mac x11 working if you’d like
<jevinskie[m]>
I was trying to get a nix formula to build a universal build a few months ago
<jevinskie[m]>
And the author recently committed some changes that look like they should bump up the rendering perf quite a bit along with trackpad support (no middle scroll wheel with a trackpad was a bad experience in older builds)
<jevinskie[m]>
GTK3 builds perf had some big regressions compared to GTK2 builds but you lost stuff like anti-aliasing
<josuah>
impressed by your patience face to this much narcissism!
<whitequark[cis]>
I don't think that's narcissism
<josuah>
maybe I just misinterpreted then
<whitequark[cis]>
I think so, yeah
<whitequark[cis]>
it's similar to the confusion I've seen people face without this much of a language barrier, but compounded by the latter
<whitequark[cis]>
ultimately this will be fixed when we have a proper first-party tutorial; they did read the docs that are there, but didn't find the answer
<josuah>
maybe me being too much marked by past interaction with schools like 42.fr where he is from
<whitequark[cis]>
42.fr?
<josuah>
A school without any price, with very well reputation, and to which you eat for free... If you are among the top $X on the listing
<whitequark[cis]>
I see
<josuah>
so it is emulated competition, with great projects nonetheless, and gerat people too, but does not encourages collaboration
<whitequark[cis]>
how'd you learn they are from 42.fr?
<josuah>
s/very well/very good/ hah looks like it's hitting me too :)
<josuah>
I could guess from 2nd message (actually thought he was from Epitech, from the same group)
<mcc111[m]>
<jevinskie[m]> "BTW mcc11 gtkwave this month did..." <- That would be interesting. Alternately, if you can tell me which version to build (IE rewind to 110 instead of 116, for example) that would work too.