Current Server Time: 16:42 (Central Europe)

#apertus IRC Channel Logs

2021/05/24

Timezone: UTC


00:24
comradekingu
left the channel
03:47
Bertl
off to bed now ... have a good one everyone!
03:47
Bertl
changed nick to: Bertl_zZ
07:28
kbeckmann
left the channel
08:31
eppisai[m]
good morning, folks! :)
09:00
anu3jn
left the channel
09:10
se6ast1an
good day
12:07
comradekingu
joined the channel
12:59
Bertl_zZ
changed nick to: Bertl
13:00
Bertl
morning folks!
13:20
Dest123
joined the channel
13:48
manav
joined the channel
14:42
aombk3
joined the channel
14:45
aombk2
left the channel
15:13
BAndiT1983|away
left the channel
15:13
BAndiT1983|away
joined the channel
15:55
bluez
joined the channel
16:00
se6ast1an
MEETING TIME, welcome everyone!
16:00
se6ast1an
who is here?
16:00
vup
is here
16:00
eppisai[m]
is here!
16:00
Bertl
is here ...
16:00
anuejn
is here
16:01
tpw_rules
is here
16:01
se6ast1an
Great!
16:02
se6ast1an
So in general it is expected of all team members and GSoC students to participate at these weekly meeting, I will also send a reminder email to all other students who did not pick up the habbit yet
16:02
Dest123
Dest123 is here
16:02
vnksnkr
is here
16:03
se6ast1an
if you cant make it let us know in advance, if you supply a brief summary of your last week activity I can share that with others in your absence
16:03
se6ast1an
as suggested by vup I would like to also discuss with students/mentors about a separate irc channel just for gsoc
16:04
se6ast1an
we used #gsoc2020 last year but to not create a new one every year what about #apertus-gsoc?
16:04
tpw_rules
what goes on in that channel specifically? administrative stuff?
16:04
vup
works for me, I guess we want to do it on libera?
16:04
tpw_rules
^
16:04
se6ast1an
yes libera brings us to the next big topic
16:04
tpw_rules
(i will be on libera later this week)
16:04
bluez
\me is here
16:04
bluez
is here
16:04
anuejn
votes also in favr of ibera
16:05
se6ast1an
tpw_rules: everything gsoc related that for some reason participants are not comfortable doing here on main chan
16:05
se6ast1an
can be administrative stuff or deep task related technical stuff between mentor/student
16:05
se6ast1an
in any case you can discuss all that here, just if you prefer not to thats the channel then for it #apertus-gsoc
16:05
se6ast1an
so libera:
16:06
se6ast1an
in case you have not seen it yet:
16:06
se6ast1an
https://hackaday.com/2021/05/20/freenode-debacle-prompts-staff-exodus-new-network/
16:06
metal_dent[m]
will move to libera soon
16:06
se6ast1an
https://fuchsnet.ch/freenode-resign-letter.txt
16:06
se6ast1an
freenode is soon to be abandoned :)
16:06
se6ast1an
we already have moved #apertus to libera.chat server
16:07
anuejn
but, I would also ask everyone to do things that can be done in the main channel - do them in the main channel
16:07
anuejn
as it is logged and can be searched afterwards
16:07
se6ast1an
currently we operate in both "worlds"
16:07
se6ast1an
but its just a matter of time when we will move there entirely
16:07
Bertl
let's make the move quickly
16:08
se6ast1an
there is no rush/urgency but also no reason to stay here longer than necessary
16:08
Bertl
we can leave a topic and maybe some log bot behind
16:08
se6ast1an
so everyone should register accounts/nicks there already and join the respective channels, there is no reason not to at this point :)
16:08
danieeel
it would be good to forward everybody to the new place, so just make a hard switch, to avoid duplicate rooms
16:09
danieeel
something like lock the chan for posting, and autoreply by a bot?
16:09
danieeel
changed nick to: danieel
16:09
tpw_rules
i have seen rumors that the new freenode staff are willing to take over migrated channels on this network, so it would be good to leave something here
16:09
se6ast1an
yes, a quick switch would be prefered but with almost 50 nicks in here I also dont want to leave people behind
16:10
vup
To help agains splitting up the channels we probably just want to set +m on the freenode channels soon
16:10
danieel
a copy/paste sync bot? :)
16:11
se6ast1an
we can target to hold the next week meeting on libera already
16:11
se6ast1an
I think a week would be a good timeframe to reach most people
16:11
se6ast1an
another topic raised recently is a separate GSoC meeting after the normal meeting
16:12
se6ast1an
where students/mentors can interact as desired
16:12
se6ast1an
where the normal meeting should focus on brief reports/next steps from everyone
16:12
se6ast1an
if thats alright for mentors students I would like to also make that a habbit
16:12
se6ast1an
1h after normal meeting
16:13
se6ast1an
UTC 17:00 that is every monday
16:13
tpw_rules
i like the idea. is it necessary to have it at its own time?
16:13
se6ast1an
not really as everyone is around anyways in the best case
16:14
se6ast1an
but should still help as a general guideline
16:14
Bertl
so let's just call it 'after the regular meeting' :)
16:14
se6ast1an
also fine :)
16:15
se6ast1an
ok quick round of updates from everyone
16:15
se6ast1an
vup any news from you?
16:15
vup
sure
16:17
se6ast1an
please share :)
16:17
vup
yep, trying to :)
16:17
vup
just crashed my browser by pasting a link into it :)
16:17
vup
Ok
16:18
vup
So I mostly worked on two things, I continued working on creating a prototype of the RPC protocol for the new control daemon.
16:18
vup
This is not yet finished, but I think I have most of the basic concepts now thought out and should have a working prototype soon.
16:19
vup
Furthermore discussions with anuejn about the current state of the nmigen gateware brought me to think about that more again aswell.
16:20
vup
Currently there are some rather magic and or unpleasent parts of it.
16:20
vup
There are mostly summarized here: https://github.com/apertus-open-source-cinema/naps/discussions/11
16:21
vup
The first thing I decided to tackle was the elaboration order problem.
16:21
vup
This occurs in a couple of places, but we can take the integrated logic analyzer as an example:
16:22
vup
In nmigen module can be composed of submodules.
16:22
vup
And a module is turned into its equivalent statements by the `elaborate` method.
16:23
vup
For the integrated logic analyzer we want to be able to add probes all over the design and then in the end connect them up to some trigger logic, sample memory and readout logic.
16:23
vup
So for this to work well, the ILA would have to be elaborated after all other modules.
16:23
vup
(Because all other modules could in theory add probes)
16:24
vup
But because nmigen will elaborate submodules only after the main module is elaborated this is very hard.
16:24
vup
So the currently idea is to create a thin layer above nmigen, which has a two phase elaboration
16:25
vup
The first phase is like the normal nmigen elaborate method, where you can add statements, submodules, etc.
16:25
vup
But additionally there is a second phase which is called after all the submodules are (recursively) elaborated, where one can add additional statements.
16:25
vup
This makes adding something like the integrated logic analyzer quite easy.
16:25
vup
You can find a current design prototype here: https://codimd.niemo.de/6yhlGdaNT8e2F9gfIPuHeQ?view
16:27
vup
The goal is, to further extend this prototype to adequately address all the pain points linked above, you can find some general ideas and notes below the code in the linked prototype
16:27
vup
Sorry for this wall of text :)
16:27
vup
Thats it from me this week.
16:28
se6ast1an
many thanks, very interesting but mostly hard to understand from an outside perspective
16:28
se6ast1an
or at least from me :)
16:29
se6ast1an
eppisai[m]: your up next
16:29
vup
Yeah, sorry, its probably easier to understand if you are atleast a bit familiar with nmigen (like tpw_rules or Bertl :)
16:29
eppisai[m]
Hello Everyone!
16:30
eppisai[m]
So, last week I declared a transition framebuffer in linker script and was analysing the linker script as to why their where two different framebuffer memory segment in kseg0 and kseg1
16:32
eppisai[m]
upon discussing with BAandiT1983, bootloader framebuffer is not being used currently and was mostly declared if wanted to added transition in future..
16:32
eppisai[m]
So have been doing that..
16:32
eppisai[m]
and found a new version of xc32
16:32
BAndiT1983
as bootloader framebuffer is not used, it won't be required there, only UART or USB connection, so no need to bother with such details
16:33
eppisai[m]
oh.. was wondering.. since framebuffer manipulations are done in firmware itself
16:33
eppisai[m]
will keep this mind :)
16:33
eppisai[m]
so, continuing..
16:33
BAndiT1983
bootloader will take the role of BIOS and data transfer
16:35
eppisai[m]
had just adjusted the current makefile for xc32 v3.00 and read about the new features.. seems very exciting (gcc now 8.3.1) that means c++11 is supported
16:35
eppisai[m]
BAndiT1983: ohk..
16:35
eppisai[m]
so, thats it..
16:35
eppisai[m]
Thanks :)
16:36
eppisai[m]
https://ww1.microchip.com/downloads/en/DeviceDoc/xc32-v3.00-full-install-release-notes.html
16:36
se6ast1an
many thanks, the changes for updated compiler have already been merged in github repo
16:36
se6ast1an
anuejn: please share your progress update next
16:36
BAndiT1983
but it still requires testing on real hardware!
16:37
anuejn
I worked mainly on the nmigen gateware since I last reported.
16:37
anuejn
While trying to write a mipi-dsi transmitter I really missed some debugging infrastructure.
16:37
anuejn
So I wrote an integrated-logic-analyzer (ILA) and some nescessary support infrastructure.
16:37
eppisai[m]
BAndiT1983: yes, will do so after todays meet. :)
16:37
anuejn
* It is now possible to communicate constants from bitstream generation time to runtime (eg. configuration).
16:38
Bertl
BAndiT1983: speaking of hardware, we should check if the EF can be used for the remote setup or alternatively we have to replace that
16:38
anuejn
* There Is now a Memory that can map to block ram on the fpga and can be read / written to by the python driver code. (in addition to the already existing CSRs)
16:38
anuejn
* There are cores for generating / reading PacketizedStreams from / to the driver
16:39
anuejn
* The ILA itself finally has many handy features like pretrigger, support for decoding FSM states to strings and writing VCD files.
16:39
anuejn
Also I renamed the 'nmigen-gateware' repo to 'naps' to be the same as the python package.
16:39
anuejn
Thats it from my side :)
16:40
se6ast1an
lots of progress, great
16:40
se6ast1an
many thanks
16:40
se6ast1an
Dest123: your turn please
16:41
Dest123
I started working on the CMV12K output channels this week
16:41
Dest123
I'm currently working on the data training over the channels
16:42
Dest123
that's it for me :)
16:43
se6ast1an
many thanks!
16:43
se6ast1an
vnksnkr: any news from your side?
16:44
vnksnkr
I've been mostly reading a few datasheets as well as the wiki this week
16:44
vnksnkr
I did get started with a pulse generator for sending the signals to the Remote.
16:45
vnksnkr
That's it nothing else from my side :)
16:45
se6ast1an
sending signals to the remote, please elaborate
16:46
vnksnkr
sure so I need to send signals to simulate the button presses on the Remote for testing
16:48
se6ast1an
right, just had to read up on your task to remind myself of the underlying hardware: https://lab.apertus.org/T1233
16:48
vnksnkr
So one challenge is to simulate the bouncing patterns and I've been working on a pulse generator that would generate random pulses within a time interval
16:48
se6ast1an
the hardware is already in place in the remote access setup I assume Bertl ?
16:49
Bertl
yes, but it has had only basic testing so far
16:50
se6ast1an
right
16:50
se6ast1an
anyone else who wants to share anything (I will share next and leave the closing words to Bertl as its tradition already)?
16:51
se6ast1an
I want to focus on getting the JTAG hardware debugger up and running with the AXIOM Remote
16:52
se6ast1an
I will need some guidance for that from Bertl but would document any findings so the nxt folks should get it setup much quicker already
16:52
se6ast1an
last week tele/factoryhub reported that the last stage of hardware production is ongoing with the soldering of the image sensors sockets
16:52
se6ast1an
everything else has been completed
16:53
se6ast1an
I did not get a heads up on it being finished yet but since today is a holiday (at least here) I will inquire again for a status update tomorrow
16:53
se6ast1an
all other boards have been received and scanned with 1200DPI for placement documentation
16:53
se6ast1an
all boards together will be handed over to Bertl as next step, hopefully this week
16:54
se6ast1an
on another front I have done pretty much everything to get the nvme ssd performance on the rock pi 4 to meaningful levels without success
16:54
se6ast1an
I get > 1 Gbit/s throughput writing zeros
16:54
se6ast1an
but for random data the maximum was around 290MB/s
16:55
se6ast1an
the same nvme ssd connected to a fast laptop over USB3 achieved ~400MB/s in exactly the same benchmark
16:55
se6ast1an
maybe you have seen the email today from Charlie from https://andahammer.com/
16:56
se6ast1an
I hope he can provide some new input with similar hardware (NanoPi M4V2)
16:56
se6ast1an
combined with nvme ssd
16:56
se6ast1an
if anyone of you have a nvme ssd lying around and any SBC with nvme ssd compatiblility let me know
16:57
se6ast1an
lastly BAndiT1983 and me worked on the online 3d cad viewer in the last week with nice progress to showcase
16:57
se6ast1an
https://apertus-open-source-cinema.github.io/cad-3d-viewer/prototype/
16:58
se6ast1an
BAndiT1983: implemented screen space reflections (not perfect yet, bit blocky and distorted on some surfaces)
16:58
se6ast1an
but IMO looking great already
16:58
se6ast1an
I added a multi material assembly (labeled "Assembly Test") that you can already try online
16:59
BAndiT1983
*added, not implemented (although thinking of adding own implementation based on newer algorithms)
16:59
BAndiT1983
*meant SSR
16:59
se6ast1an
and emount and cap bottom also show a location image now that displays where this part is located in the camera assembly
16:59
Bertl
opacity looks like it controls the reflectiveness here
16:59
BAndiT1983
it's called opacity in SSRPass, so kept the name
16:59
Bertl
but it is impressive how good it loops
16:59
Bertl
*looks
17:00
se6ast1an
thanks :D
17:00
se6ast1an
next steps are a UI overhaul with screendesign here: https://paste.pics/3f0e34773ff7099ae46e98d2caaabf67
17:00
se6ast1an
in general I am not a big fan of dark UIs so thinking of the same in a bright design (just inverted colors ) : https://paste.pics/abde01cdc389d719899b3448f8719843
17:01
se6ast1an
what are your thoughts on bright vs dark ?
17:01
se6ast1an
BAndiT1983 already promised to add a switch to change themse easily
17:01
se6ast1an
but I think a general idea of peopels opinion would also be nice
17:02
Bertl
I like dark themes ;)
17:02
se6ast1an
noted :)
17:02
metal_dent[m]
Dark \o/
17:02
tpw_rules
i like light themes
17:02
eppisai[m]
I like the idea of giving user the option.. that is dark and light toggle button
17:02
bluez
dark always :)
17:03
BAndiT1983
some dark souls here
17:03
vnksnkr
:)
17:03
eppisai[m]
:D :D
17:03
se6ast1an
thanks, seems like a toggle option is indeed the best way to go :)
17:04
bluez
hehe.. yep toggle will be best i think
17:04
se6ast1an
we will continue tweaking the materials as well and the GUI
17:04
se6ast1an
I think thats it from my side
17:04
se6ast1an
Bertl: your honor?
17:04
Bertl
thanks
17:05
Bertl
well, not too much happened here this week, I mainly spent the time on getting the remote setup up and running
17:05
Bertl
we will do a session after this meeting (after a short break) for our 2021 students
17:06
Bertl
I also did some tests on the gateware side regarding splitting up the input and the output pipeline
17:07
Bertl
which seems for me the way to go in the future
17:07
Bertl
except for helping students here and there that's it for the last week
17:08
se6ast1an
many thanks!
17:08
se6ast1an
anyone else with anything to share/report/discuss?
17:09
se6ast1an
right then! MEETING CONCLUDED!
17:10
se6ast1an
feel free to start the task related discussions here after your short break
17:10
Bertl
okay, let's take a 20min break, and continue at 19:30 with the remote setup on #apertus-gsoc
17:13
Bertl
(let's use the libera.chat network for that ;)
17:13
kengz[m]
joined the channel
17:34
BAndiT1983
left the channel
17:34
BAndiT1983
joined the channel
17:37
BAndiT1983
left the channel
17:37
BAndiT1983
joined the channel
17:58
manav
left the channel
18:01
manav
joined the channel
19:31
manav
changed nick to: Manav
19:31
Manav
changed nick to: Manav_
21:51
rektide
joined the channel
21:52
rektide
left the channel
21:52
rektide
joined the channel
21:54
rektide
left the channel
21:54
rektide
joined the channel
21:55
rektide
left the channel
21:55
rektide
joined the channel
22:05
rektide
left the channel
22:05
rektide
joined the channel