Current Server Time: 16:41 (Central Europe)

#apertus IRC Channel Logs

2021/07/19

Timezone: UTC


04:13
Bertl_oO
off to bed now ... have a good one everyone!
04:13
Bertl_oO
changed nick to: Bertl_zZ
07:28
se6astian
good day
10:04
anuejn
tpw_rules: we have the new shiny fatbitstream generation :)
10:04
anuejn
including super fast rebuild caching in case of only driver changes
10:49
mixfix41
joined the channel
10:58
danieel
left the channel
11:00
danieel
joined the channel
11:16
Bertl_zZ
changed nick to: Bertl
11:16
Bertl
morning folks!
11:44
se6astian
hello!
12:47
Dest321
changed nick to: Dest123
12:48
Dest123
Good Morning!
12:59
Bertl
off for now ... bbl
12:59
Bertl
changed nick to: Bertl_oO
15:52
aombk2
joined the channel
15:56
aombk
left the channel
16:00
Bertl_oO
changed nick to: Bertl
16:00
se6astian
MEETING TIME, who is here?
16:00
Bertl
is here ...
16:00
Dest123
is here
16:00
vnksnkr
is here
16:01
tpw_rules
is here
16:01
se6astian
Dest123: would you like to start?
16:01
Dest123
yeah sure
16:01
Dest123
so last week I had the channels ready for testing
16:02
Dest123
I integrated them with the existing hardware, where some problems occurred but I managed to solve them with Bertl
16:03
Dest123
the channels passed the training phase successfully, but I encountered a problem whilst capturing an image
16:03
vup
is here
16:03
Dest123
right now I'm working on using the sequencer trigger signals which hopefully would solve the problem
16:04
Bertl
note here: the channels are connected at the parallel interface and generate a 'fixed' pattern
16:04
Dest123
yes
16:05
aombk3
joined the channel
16:05
Dest123
that's it for me this week, thank you :)
16:05
se6astian
many thanks!
16:05
se6astian
vnksnkr go ahead
16:05
vnksnkr
Hi
16:06
vnksnkr
last week i didnt have much progress..as I was facing a few issues with the remote hardware
16:07
vnksnkr
I initially was using the driver developed last year to upload my bitstream on the MachXO2
16:07
vnksnkr
I was'nt able to decode and send back any commands from the fabric
16:08
vnksnkr
and as discussed earlier in the chat, the machXO2 was being reset, when the driver was being unloaded
16:08
vnksnkr
which I should have found out earlier
16:09
aombk2
left the channel
16:09
vnksnkr
so now I've shifted to using the python scripts to load the bitstream
16:10
vnksnkr
hopefully I can get it uploaded succesfully and finalize my gateware by this week
16:11
vnksnkr
that's it from my side
16:11
Bertl
did you work with bluez to get this sorted?
16:11
vnksnkr
yes there are 2 options right now to work with the jtag interface
16:12
Bertl
I think it would be good to remove the reset from the driver unload or at least make it optional (e.g. via driver option)
16:12
Bertl
so that you can use the driver to upload gateware and then free up the pic registers
16:13
vnksnkr
ok, another option bluez suggested was to use svf files to generated the commands since the driver supports openocd
16:13
Bertl
also bluez was using/testing a python script for burst upload to the MachXO2 before he wrote the driver, which should work with the bitstream IIRC
16:14
Bertl
okay, that would be the third option then, to use the openocd interface the driver provides to send/receive the jtag data
16:15
Bertl
okay, thanks!
16:16
se6astian
right, tpw_rules is next, please go ahead
16:16
tpw_rules
alright so i have good news, i got the prototype of the link training working for the cmv12k on the beta
16:17
Bertl
yay!
16:17
tpw_rules
it's not quite finished yet, i have some improvements to the training algorithm to make still and documentation to finish up. but i hope to address those today or tomorrow and get it merged
16:17
tpw_rules
then the next step is porting Bertl's pixel remapper from verilog and we should be able to see some images from the sensor soon :D
16:17
vup
*vhdl
16:17
tpw_rules
yes, sorry
16:18
Bertl
and it is not mine ;)
16:18
tpw_rules
but it does seem to work well on anuejn's beta. i want to try it on the others when they have sensors just to make sure it is robust
16:18
tpw_rules
also i heard anuejn fixed some complains i had with the pydriver compilation speed so work should be easier now. thank you!
16:19
tpw_rules
that should be it from me this week
16:19
Bertl
looking forward to the new gateware!
16:20
se6astian
many thanks
16:20
se6astian
anuejn are you there?
16:21
se6astian
BAndiT1983 shared some news but excused himself from the meeting
16:21
se6astian
he has been working on browser based remote "emulator"
16:21
se6astian
https://i2.paste.pics/82e0a2098b5f7ce278a7532cc469abe9.png
16:22
se6astian
this is related to https://lab.apertus.org/T1178
16:22
se6astian
eppisai: are you here?
16:22
se6astian
quick updates from me: I returned from vacation just 2 days ago so not much to report
16:23
se6astian
we might be able to collaborate with the technical university vienna on CNC manufacturing on camera related metal parts
16:23
se6astian
a meeting will happen probably at the end of august when manfred is back from tokio
16:23
eppisai
hi..i don't have much progress to report
16:23
eppisai
i am having some issues with remote hardware.. so trying to work that out
16:23
se6astian
right
16:24
se6astian
anyone else who wants to report?
16:24
Bertl
eppisai: what kind of issues?
16:24
eppisai
the mini module is not working..
16:24
eppisai
:(
16:24
eppisai
its not broken.. i am sure
16:25
se6astian
did you try it on another laptop as planned?
16:25
Bertl
did you switch the cables?
16:25
eppisai
i have switched the cables, tried on another laptop.. showed nothing..
16:26
se6astian
very weird
16:27
se6astian
right Bertl please share your closing words/progress with us?
16:27
Bertl
okay
16:27
Bertl
well, last week, some of our students touched the remote hardware the very first time ;)
16:28
Bertl
(probably related to the first evaluation ...)
16:28
Bertl
and some issues were found which resulted in me reworking one of the betas to make it work as intended
16:28
Bertl
and rearranging the order so that the beta-c is nearby the remote
16:29
Bertl
I also spent some time on further testing of the Axiom Beta PCBs and will hopefully get this week to actually assembling new Betas
16:29
tpw_rules
yay
16:30
se6astian
great!
16:30
vup
nice
16:30
Bertl
the video stream is still an open issue, I investigated youtube and twitch but both require a quite complicated setup and have huge delays
16:31
tpw_rules
what do you consider "huge"?
16:31
Bertl
5-15 seconds
16:31
se6astian
what about jitsi?
16:32
Bertl
maybe an option, haven't considered it yet
16:32
vup
hmm I have certainly managed to get the yt latency to go under that
16:32
vup
also what about just a basic nginx + rtmp plugin setup?
16:32
se6astian
yt can go down to slightly below 5s
16:33
Bertl
I'm open for all solutions, the best performance/ease of use I had so far was with a server at the university of vienna
16:33
Bertl
which basically got a fixed stream connect and then re-distributed the data on a per user base
16:34
tpw_rules
i think that time is probably fine?
16:34
Bertl
this way we had about a second delay and almost no load
16:34
Bertl
unfortunately we cannot use the server for our streaming purpose
16:35
Bertl
so, I'm open to all suggestions, as long as I do not have to waste too much time on it ;)
16:36
Bertl
that's it from my side for this week
16:36
se6astian
many thanks
16:36
se6astian
anyone else who wants to report/share anything?
16:41
vup
Bertl: do you know what software they are running on that server of the university vienna?
16:42
Bertl
I should be able to figure that out, it is likely some rtsp proxy
16:43
se6astian
ok, MEETING CONCLUDED, many thanks everyone!
16:44
Bertl
thanks for the moderation!
16:44
tpw_rules
thank you!
16:45
Dest123
thank you :)
17:43
Bertl
off for now ... bbl
17:43
Bertl
changed nick to: Bertl_oO
20:13
Bertl_oO
vup: rtsp-simple-server was the software used
20:14
vup
I see
20:15
vup
and you were streaming to that server using ffmpeg?
20:15
vup
what was used on the viewing / receiving side?
20:15
vup
rtsp aswell, or rtmp or hls?
20:15
Bertl_oO
yes, it was an rtsp stream with ffmpeg
20:17
vup
so ffmpeg on both sides?
20:48
Bertl_oO
client can use whatever app they want
20:49
Bertl_oO
was tested with vlc, mpv and ffplay IIRC
21:00
vup
sure, but the client can also affect the latency, if I understand the server right, it has the ability to "transcode" from rtsp to hls for example, which will probably add some latency compared to just listening to the rtsp stream
21:02
Bertl_oO
probably
21:49
vup
Bertl_oO: did the rtsp-simple-server work behind nat?
21:50
Bertl_oO
no idea
23:10
anuejn
so here is my report (a little late to the party):
23:10
anuejn
I did some rather large rework of the pydriver generation in naps
23:11
anuejn
now we produce zip files that can be executed via python and are self unpacking
23:11
anuejn
they behave much like the last iteration where we had self extracting shell scripts with a bunch of base64 in them
23:12
anuejn
this fixes some issues with to long arguments and is generally much more hackable after building the bitstream
23:12
anuejn
also I rewrote much of the infrastructure that _generates_ the logic and runs the build process
23:13
anuejn
we now have better caching (does not rebuild the bitstream if only driver code changed)
23:14
anuejn
also if we have cache hits, the system is able to find that out earlier resulting in a much smoother development experience :)
23:14
anuejn
thats it from my side
23:18
Dest123
left the channel
23:19
Dest123
joined the channel
23:34
vup
Nice