Current Server Time: 17:27 (Central Europe)

#apertus IRC Channel Logs

2020/01/13

Timezone: UTC


00:21
Bertl_oO
off to bed now ... have a good one everyone!
00:22
Bertl_oO
changed nick to: Bertl_zZ
01:14
RexOrCine
changed nick to: RexOrCine|away
01:27
illwieckz_
joined the channel
01:31
illwieckz
left the channel
02:07
illwieckz_
changed nick to: illwieckz
04:18
BAndiT1983|away
changed nick to: BAndiT1983
04:36
BAndiT1983
changed nick to: BAndiT1983|away
09:51
Bertl_zZ
changed nick to: Bertl
09:51
Bertl
morning folks!
12:24
illwieckz
left the channel
12:25
illwieckz
joined the channel
13:27
illwieckz_
joined the channel
13:31
illwieckz
left the channel
14:22
BAndiT1983|away
changed nick to: BAndiT1983
14:33
comradekingu1
joined the channel
14:33
comradekingu
left the channel
16:01
se6ast1an
meeting time!
16:02
se6ast1an
who is present?
16:02
anuejn
is present :)
16:02
Bertl
is present.
16:02
vup
is present aswell ;)
16:03
se6ast1an
hurray ;)
16:03
se6ast1an
right, last meeting was quite some time ago
16:03
se6ast1an
possibly mid december, cant even remember anymore
16:03
se6ast1an
so I would suggest we give wider reaching updates of what happened recently to get everyone on the same page
16:04
se6ast1an
who wants to start?
16:05
se6ast1an
vup maybe you can give a short report from CCC?
16:05
vup
sure
16:06
vup
so we had a assembly at 36C3 like the last years and setup a 1-2 beta's and a micro
16:06
se6ast1an
felix_: you here too?
16:07
vup
one beta was supposed to demo the sdi support / plugin module, but unfortunately there were some mechanical problems with different connections, meaning that it wasn't working most of the time
16:07
se6ast1an
oh no
16:08
felix_
se6ast1an: yep
16:08
vup
the beta hdmi demo and the micro demo were working fine however and like usual attracted quite some crowd interested in the project
16:09
vup
i guess that would be it; any questions?
16:10
Bertl
what is the typical HDMI demo?
16:10
se6ast1an
great, any contacts worth following up? did you meet drew from oshpark?
16:10
vup
Bertl: connecting the beta to a monitor using the HDMI plugin module
16:11
Bertl
okay, so standard live HDMI feed then
16:12
vup
se6ast1an: lots wanted to reach out after the event, but nothing concrete; don't think drew was at the assembly (but I also wasn't there all the time)
16:12
se6ast1an
right, thanks
16:12
felix_
was most of the time at the open source firmware assembly; formerly known as the coreboot assembly
16:14
se6ast1an
felix_: what are your plans regarding SDI for 2020?
16:15
vup
ah and of course the new webui / control daemon integration also proved handy for changing gain / exposure on the fly from my phone
16:15
felix_
it seems that i might have the rest of the week to continue working on photosdi
16:16
se6ast1an
sounds good, and beyond this week?
16:17
se6ast1an
webgui/nctrl <- great!
16:17
felix_
can't really plan for this year yet; currently waiting on a contract for a job
16:17
se6ast1an
right
16:17
se6ast1an
vup/anuejn what are your plans for 2020?
16:18
anuejn
thats a large question ;)
16:18
Bertl
survive?
16:19
anuejn
but since i am done with school, there might be quite some time for hacking after my exams :)
16:19
anuejn
altought less time before :(
16:21
se6ast1an
what are the plans for the axiom micro?
16:22
vup
for my part: first finishing up some control daemon work, mainly
16:22
vup
- changing some internal structures to allow handling things like large luts better
16:22
vup
- actually handling large luts better
16:22
vup
- replace / add a new API to the control daemon that for example allows clients to get notified when values change
16:22
vup
- move more of the current script stuff to the control daemon
16:22
vup
After that I have no concrete plans, but possible things include
16:22
vup
- working on the micro rev 3
16:22
vup
- continuing the work on a new approach to the gateware (using nmigen and python abstractions)
16:22
se6ast1an
anuejn: school ends this semster? not in summer?
16:22
vup
But until about mid febuary I am also probably quite limited in the time i have
16:22
vup
(sorry for the tabs)
16:22
vup
se6ast1an: concrete things that need to be done before actual work on the micro rev 3 are imo:
16:23
vup
- getting hdmi to work or proving that the current pcb is not capable of that
16:23
vup
- testing some ideas we had for the connection of a replacable image sensor
16:24
vup
s/replacable/interchangeable/
16:24
Bertl
HDMI is planned via the HDMI plugin module?
16:24
anuejn
se6ast1an: beginning of june
16:24
anuejn
Bertl: yes
16:25
anuejn
we already made 1 failed attempt to get it working
16:25
Bertl
so that should be easy to test with one of the HDMI examples, no?
16:25
anuejn
but didnt really debug anymore
16:25
anuejn
(we used litehdmi)
16:26
vup
the main problem with HDMI is that currently we are not getting past the HDMI transceiver on the plugin module
16:26
Bertl
well, would probably make sense to test with 'known working' code first
16:26
vup
we get a valid looking signal from the plugin connector but the HDMI transceiver just outputs nothing
16:27
vup
Bertl: what 'known working' code would you suggest?
16:27
Bertl
there should be an HDMI test code which just outputs a test image
16:27
Bertl
(have to dig it out and probably update it to recent vivado)
16:28
Bertl
you also want to make sure that the GPIOs are set correctly, but that is something static
16:28
Bertl
anyway, we can discuss that later ... just remind me\
16:28
vup
well if you can dig that up i would be happy to test it
16:29
se6ast1an
great, many thanks
16:29
se6ast1an
anything else vup/anuejn? otherwise Bertl can start
16:30
anuejn
i dont think so
16:30
vup
thats it from me i think
16:31
vup
ah also i think https://github.com/apertus-open-source-cinema/axiom-beta-firmware/pull/127 should be ready to merge now; but we can discuss that later aswell
16:31
se6ast1an
great
16:31
se6ast1an
Bertl?
16:31
Bertl
okay, so I'll cover the two or three most important developments over the last month or so as I have no clue what we discussed in the previous IRC meetings
16:32
Bertl
basically during GSoC 2019 while Apurva was working on the USB3 Plugin Gearwork, we discovered, that the FTDI chip used in the hardware is a little problematic
16:33
Bertl
we knew already that we won't be able to get full USB 3.0 bandwidth with this solution, because the parallel interface on this chip is not fast enough
16:34
Bertl
nevertheless, we expected to get about 400MByte/s throughput in one direction (camera to PC)
16:35
Bertl
various tests showed that this doesn't seem to be true unless there are really special conditions met, like for example completely clean/free USB3 bus, high end PC, perfect host controller, etc
16:36
Bertl
and even there the best we reached so far was about 380MByte/s
16:36
Bertl
naturally we started investigating other alternatives and solutions to mitigate this
16:37
se6ast1an
did you try the different usb3 controller cards already for this ?
16:37
Bertl
not yet
16:38
Bertl
one very promising but rather long term plan is to use a 10G ethernet serdes to transfer up to 12Gbit/s of data between the AXIOM Beta and a PC
16:39
Bertl
this uses a special chip and is planned to work over a USB-C cable (not talking USB :)
16:40
Bertl
as I mentioned, this is rather long term, as we need a new plugin and a special PC card for this to work, but the first steps (prototype design) have been made and we will assemble a test setup in the next few weeks
16:41
anuejn
Bertl: do you think that this is a problem intrinsic to usb3 or just affects the ftdi?
16:41
Bertl
to be honest, not completely sure there
16:41
Bertl
on paper, USB 3.0 should be capable of 5Gbit/s
16:43
Bertl
i.e. 625MByte/s so with the 8/10 encoding and protocol overhead, that should leave about 480MByte/s throughput
16:44
Bertl
knowing that we will max out at 400MByte/s with the parallel interface, we should have plenty of room
16:44
Bertl
one known issue with the FTDI is that it doesn't support isochronous transfers at all
16:44
Bertl
i.e. bandwidth reservation is not an option with this hardware
16:45
Bertl
a more short term solution to the problem is replacing the FTDI with a Cypress FX3 chip
16:46
Bertl
which is more flexible (you get access to the ARM core there), supports isochronous transfers and all kind of other useful stuff, has probably (not verified yet) more bandwidth on the parallel side
16:46
Bertl
but also requires more software to function
16:47
Bertl
finally the third option here is to use HDMI to transfer raw data between the camera and PC
16:48
Bertl
we already investigated this with the experimental 4K raw modes and it worked reasonably well, except for the unavoidable encoding on the recorders
16:49
Bertl
searching for usable solutions in this direction, we stumbled upon various game capture cards and solutions and also a special HDMI capture card which is USB based
16:50
Bertl
the CamLink 4K made by Elgato a former German company now owned by Corsair
16:51
Bertl
this adapter is designed to connect your camera (with HDMI out) to your PC and allows to capture up to 4K (although only 30FPS and special conditions :)
16:51
Bertl
we also found a good deal before christmas on those so we ordered one to see how it does
16:52
Bertl
while the off-the-shelf performance was not overwhelming (but it is okay)
16:52
Bertl
opening up the device gave a bunch of pleasant surprises ...
16:53
Bertl
first, they use the FX3 chip for the USB3 interface, then they use a Lattice ECP5 as gearwork and finally an iTE HDMI receiver capable of 4K as input
16:55
Bertl
somehow this information triggered a reverse engineering avalanche on the internet, and over a few weeks basically everything relevant was discovered
16:55
Bertl
we now have tools to reprogram the CamLink devices and the FPGA hardware has already been running Linux on LiteX
16:56
Bertl
so it's basically down to writing FX3 firmware and ECP5 gateware to use it as HDMI to USB solution with the AXIOM Camera
16:57
Bertl
and when we control both ends, we can easily use this to transfer arbitrary data
16:58
Bertl
that's basically the current state ... did I miss anything?
16:58
Bertl
Ah, yes, one important part I missed
16:59
se6ast1an
"FX3 firmware and ECP5 gateware" is what you are working on next right?
16:59
Bertl
we also adapted the AXIOM Beta Power Board for the ABCP by removing the FTDI/JTAG interface and adding DC/DC converter to allow for higher input voltage and battery operation as well as power control
17:00
Bertl
the designs have been submitted to oshpark and will probably arrive this or next week for a prototype assembly
17:01
Bertl
@se6ast1an: yes, I'm currently playing around with the FX3 devkit to figure out the limits and get a feeling for the firmware there
17:02
Bertl
I guess that's it from my side.
17:02
Bertl
https://wiki.apertus.org/index.php/Elgato_CAM_LINK_4K
17:02
se6ast1an
"allow for higher input voltage and battery operation as well as power control" <- that means we can power higher than 5V fans inside the CP enclosure
17:02
se6ast1an
and the power source can be a battery, what range did you say we have 6-12 V ?
17:03
Bertl
between 6V and 17V
17:03
se6ast1an
it will means we need to use 12V power bricks in the future instead of 5V ones and USB batteries will not be useable anymore in the future
17:04
Bertl
well, the new Power Board still has the dual 5V inputs as well
17:04
se6ast1an
but any lithium-ion based battery: 2x3.7V typically
17:04
se6ast1an
ah!
17:04
se6ast1an
so its one or the other?
17:04
Bertl
so if you want to use those, they are still available, but the power control will be disabled for those
17:04
se6ast1an
so no on/off button?
17:04
se6ast1an
for 5V?
17:05
Bertl
no system controlled on/off for direct 5V operations, yes
17:05
Bertl
*operation
17:08
se6ast1an
right
17:08
se6ast1an
quick update from me then
17:08
se6ast1an
gsoc 2020 applications open tomorrow
17:09
se6ast1an
I will be at FOSDEM at the end of the month and present the project/beta there
17:11
se6ast1an
new compact cnc milled prototype parts should arrive in a week or so
17:11
se6ast1an
and I have been and am working right now on the AXIOM Remote c++ port
17:11
se6ast1an
this was the new year animation greeting: https://twitter.com/ApertusOSCinema/status/1212177166332579840 :)
17:12
se6ast1an
thats it from my side
17:12
se6ast1an
BAndiT1983: anything to report?
17:12
BAndiT1983
no
17:12
se6ast1an
right, meeting concluded then, until next week!
17:19
vup
Bertl: i am gonna grab something to eat, but after that i would be open to test the "known working" hdmi code, if you can find it ;)
17:20
Bertl
okay
17:31
illwieckz_
changed nick to: illwieckz
20:01
RexOrCine|away
changed nick to: RexOrCine
20:09
BAndiT1983
changed nick to: BAndiT1983|away
20:10
BAndiT1983|away
changed nick to: BAndiT1983
20:12
frieder[m]
left the channel
20:12
pedro[m]1
left the channel
20:12
toralf[m]
left the channel
21:06
BAndiT1983
changed nick to: BAndiT1983|away