<wolfspraul>
wpwrak: can you draw a (partial) schematic of the solution with logic gates that you proposed for the m1 bootup fix?
<wolfspraul>
I'm asking because we may want to expedite rc4 so if we can parallelize some of that work it might be helpful. And you may still have the idea clearly in your mind now :-)
<wpwrak>
sure. that idea is very clear ;-)
<xiangfu>
larsc, Hi where is the milkymist linux/openwrt source code URL?
<wpwrak>
lekernel: by the way, i was thinking of a cool application for video input: if you have "trigger points" in the camera image where a movement/presence of an object can cause the generation of visual effects (you once said this would be possible), you could also do the same with audio effects. that way, one could "play an instrument" by, say, moving one's hand into the right area
<lekernel>
yeah right... but so many things so little time, heh
<wpwrak>
lekernel: now, to do this properly, you'd probably want to send this event to a MIDI instrument. that may be a PC. for this, you would need a USB-to-MIDI dongle. i looked around and they seem to come in a very large price range, around USD 5 to USD 50. would you generally expect them all to work with the M1 ? or would a prospective buyer need some guidance ?
<lekernel>
there's a proper audio out on the M1
<wpwrak>
(little time) yeah, i know :) but that's now why i mention it. it led me to the MIDI issue.
<larsc>
are usb midi devices standardized?
<wpwrak>
(audio out) that could be an alternative in some cases, yes. how good are the M1's audio capabilities, compared to the average synthesizer (or let's say qsynth, for something that's all free software)
<lekernel>
as good as you want, it's a normal ac97 audio chip
<wolfspraul>
and audio noise was reduced by 90% from rc2 to rc3! :-)
<wpwrak>
(regarding midi) it's a pity that M1 doens't have a USB device port (not counting the JTAG dongle here). otherwise, MIDI-over-USB (etc.) could be added later.
<wpwrak>
hehe ;-)
<wolfspraul>
switched from a National LM4550B to a Wolfson WM9707 codec
<wpwrak>
wolfspraul: i'm still a bit scared that adam can actually measure it so easily. but maybe what he really sees are measurement artefacts :)
<lekernel>
you can have USB device with a passive adapter, but expect major pains from the usb protocol
<wolfspraul>
wpwrak: that may well be, but I think we can safely say it's "much reduced"
<wpwrak>
(ADV7181C) oh, sounds like a big change ?
<wolfspraul>
no supposedly a very small change, drop-in, plus remove some filter outside
<wpwrak>
ah, getting rid of filters is always nice
<wolfspraul>
in that list, it sounds like the biggest change would be the logic gates, that's why I asked whether you can help with a drawing for clarity... but that's not urgent, Adam is deep in rc3 now and there seem to be some nasty problems at the moment that need to be investigated
<wolfspraul>
I'm just trying to help with some of the rc4 preparations to offload from Adam
<wolfspraul>
yes it sounds like the change from adv7181b to c is less risky than the one from National to Wolfson audio codec we did between rc2 and rc3
<wolfspraul>
I guess it's similar, even the same manufacturere here... what can possibly go wrong, right? :-)
<wpwrak>
always the same ;)
<wolfspraul>
wpwrak: I was trying to understand your midi idea above. you basically want to turn m1 into a midi emitting instrument, and then the usb2midi thing.
<wpwrak>
yes
<wolfspraul>
but m1 already has midi output, why not just use that?
<wpwrak>
that's the idea. the question was whether the usb-to-midi dongles should all work with M1 or if some selection is necessary
<wolfspraul>
why do you need such a dongle? why not just run a midi cable from the midi-out of m1 to whereever you want to run it to?
<wpwrak>
i also found it surprising that there's quite a large price range from these, from USD 5 to 50, without considering "fancy" brands
<wolfspraul>
could be just marketing & channel power, don't underestimate that
<wpwrak>
the "wherever" may be your PC. does yours have a MIDI plug ? ;-)
<wolfspraul>
we always look at things from an engineering perspective, but that be quite off economically :-)
<wolfspraul>
ah ok, so it's a software problem then?
<wpwrak>
sure. could be just that. just want to know what the expectation is :)
<wolfspraul>
I still don't understand. if you run a usb cable to your computer, why need a dongle?
<wolfspraul>
what does the dongle do?
<wolfspraul>
do you have a url for the dongle? I think I just don't get what it does...
<wpwrak>
the dongle translates from MIDI-over-USB (which is what your PC can speak) to MIDI-sans-USB (which is what M1 speaks)
<GitHub10>
[linux-milkymist/master] Merge commit 'v3.0' into lm32 - Lars-Peter Clausen
<wolfspraul>
wpwrak: btw, that example of the 50 USD cable eating into our margin is why we need to have our own accessories source anyway, so we can quickly turn around and sell them our own, 'Milkymist MIDI cable', also for 50 USD retail
<wolfspraul>
with nicely designed packaging etc.
<wolfspraul>
retail will love it
<wolfspraul>
once we get to that point I definitely need to get Steve's advice and experience back into the team, I don't want to guess those decisions
<wolfspraul>
he has been doing this kind of pushing and pulling over who eats into which margin for 10+ years
<wpwrak>
(selling accessories) yeah, that's where the big money is :)
<larsc>
will there be milkymist ethernet cables with gold pins?
<wpwrak>
quick, who will make the benchmark that shows the gilded cable has less packet loss ? ;-)
<lekernel>
packet loss? better talk about (subjective) audio quality
<lekernel>
Also, please make sure you have the following patch applied (at the time of writing this email, the patch hasn't been applied to the qemu master branch):