Current Server Time: 01:41 (Central Europe)

#apertus IRC Channel Logs

2015/10/03

Timezone: UTC


00:13
irieger
Ah, thx. Thats the image I saw multiple times but wasn't finding again ...
00:18
irieger
Ok. So all relevant for me is in FPGA. Nice ...
00:19
Bertl
yup, the arm cores are just doing configuration and setup
00:19
irieger
I'd be fine with C, CUDA, OpenCL etc. but this vhd stuff is totally new ...
00:20
irieger
I knew how FPGAs work but haven't thought about it till now ...
00:20
irieger
Meaning that it isn't programming in the classical terms
00:20
Bertl
it isn't that complicated, you just have to let go of the "normal" programming concepts
00:21
irieger
Sounds good. Have too look at this soon.
00:23
irieger
I see that LUTs are used for linearization and gamma in the flow chart. Would this be generally possible with formulas instead of LUTs or is this likely more performance-hitting on the FPGA? (I know this is kind of a bit generalized)
00:25
irieger
Do you have a good online resource for VHD/FPGA learning?
00:26
irieger
Btw. anyone here familiar with board construction willing to build a 4K LUT Box? Would be a nice project for some open hardware. There isn't even one of the commercial lut box vendors offering one for 4K ...
00:26
fsteinel
left the channel
00:26
fsteinel_
joined the channel
00:27
Bertl
there are a huge number of courses and introductions online, but the really good stuff is in (mostly older) books
00:27
irieger
You know some good ones? Maybe we have something in our library at university.
00:28
irieger
Best thing there is that more and more books are even available online for free. Some even as full book PDF files, maybe I'm lucky
00:28
Bertl
VHDL programming by example was nice to read (Perry, IIRC)
00:30
irieger
Can't find this one but there are at least 9 ebooks about VHDL
00:30
Bertl
then the Designer's Guide to VHDL
00:31
irieger
that one is available
00:31
Bertl
ah, I found one online resource (freely available)
00:31
Bertl
http://freerangefactory.org/books_tuts.html
00:31
Bertl
Free Range VHDL
00:32
Bertl
for coding techniques, there is also Essential VHDL: RTL Synthesis Done Right
00:34
Bertl
what do you consider a 4k LUT box?
00:35
irieger
Would like to have one to drive my grading monitor. It's a "cheap" one which needs a LUT to get some decent results. Would like to have a LUT box for 4K monitoring to make full use of my system here
00:36
Bertl
define LUT :)
00:36
Bertl
also, what is the interface to the monitor
00:38
irieger
Input HDMI or preferably SDI, output HDMI or display port for this case or preferably SDI so it can be used before the SDI->HDMI converter Box. Two outputs would be fine. One pass through, one with LUT applied ;-)
00:39
Bertl
so it should also work as protocol converter then
00:39
irieger
LUT = at least 33x33x33, better 65x65x65 sized 3D lookup tables ;-)
00:39
irieger
*getting greedy* ;-)
00:39
Bertl
so also 3D luts with interpolation :)
00:39
irieger
Bertl: no, pure SDI would be better I'd say
00:40
irieger
yeah, wouldn't really work without I think ;-)
00:41
irieger
I'd like to try one of the many 4K recorders/monitors like Atomos, Video devices and so on to see how good they would work and how precise they'd be.
00:41
Bertl
and you are looking for somebody to build the hardware or hardware and software?
00:42
irieger
We have a PIX-E5 at our universityso maybe I can try with it
00:43
irieger
Depending on the complexity of the software needed (which depends on the hardware I think) ;-) But wouldn't really have time for such a project even if it would be nice.
00:56
Bertl
well, if I hear from somebody interested in doing that, I'll let you know
01:09
irieger
Ok
01:11
irieger
good night
01:11
Bertl
nn
02:46
Bertl
off to bed now ... have a good one everyone!
02:46
Bertl
changed nick to: Bertl_zZ
03:51
comradekingu
left the channel
07:36
mithro
irieger: it might be possible to do what you want with the Opsis board
07:36
mithro
irieger: it would be pushing it to the limit, but it could possibly work
07:47
Bertl_zZ
changed nick to: Bertl
07:47
Bertl
morning folks!
07:49
Bertl
mithro: 33^3 3D lookup tables with interpolation is probably outside the range of the opsis, especially at 4k speeds ... but I would love to see something like that implemented on it.
07:50
mithro
Bertl: 33x33x33 isn't bad, 33bit x 33bit x 33bit is :P
07:50
Bertl
I think what irieger meant was 33 point per color channel
07:51
Bertl
so you need 33^3 * 12 or 16 bits memory
07:51
Bertl
plus some kind of interpolation between 2-3 points fetched per channel
07:53
mithro
Bertl: hrm - 33^3 * 12bits is only ~64kbytes according to wolfram alpha?
07:54
Bertl
yes, but you need to look up at least 6 values in those 64k for each channel at ~300MHz
07:55
Bertl
and then interpolate between those to form the results
07:55
Bertl
but yeah, maybe the Opsis can do that, haven't checked yet, you probably know better than I
07:56
mithro
Bertl: it would definitely be tough
07:56
mithro
Bertl: You are going to use up a lot of your FPGA space replicating the lerp block so you can do a bunch of pixels at the time
07:57
mithro
Bertl: all your block ram you'll need to use as a cache for the table data stored in the DDR ram
07:58
mithro
Bertl: getting it all to fit is going to be a constant game of trade offs
09:52
se6astian|away
changed nick to: se6astian
09:53
ShinyCyril
joined the channel
09:53
ShinyCyril_
left the channel
09:55
se6astian
changed nick to: se6astian|away
10:39
intracube
left the channel
12:11
intracube
joined the channel
12:22
niemand
joined the channel
12:55
Bertl
off for now ... bbl
12:55
Bertl
changed nick to: Bertl_oO
13:01
intracube
left the channel
13:11
intracube
joined the channel
13:40
ItsMeLenny
joined the channel
14:06
ItsMeLenny
left the channel
14:11
comradekingu
joined the channel
15:40
se6astian|away
changed nick to: se6astian
16:47
niemand
left the channel
17:41
comradekingu
left the channel
18:42
comradekingu
joined the channel
19:56
niemand
joined the channel
20:02
niemand
left the channel
20:23
niemand
joined the channel
20:55
niemand
left the channel
20:59
gcolburn
joined the channel
21:59
gcolburn
left the channel
22:11
se6astian
changed nick to: se6astian|away
23:00
antiatom
joined the channel