| 01:47 | jucar | joined the channel |
| 01:52 | jucar1 | joined the channel |
| 07:34 | s3bastian | joined the channel |
| 07:41 | dmj_nova | hi s3bastian
|
| 07:52 | s3bastian | good morning ;)
|
| 08:43 | Bertl | morning everyone!
|
| 09:00 | s3bastian | hello! why up so early? :)
|
| 09:20 | Bertl | don't ask! :)
|
| 09:33 | Bertl | dmj_nova: back with the zedboard yet?
|
| 09:37 | Bertl | s3bastian: btw, during testing the axi stuff, I had massive problems with the debug interface losing connection to the DAP, which seems to be 'expected' when the arm cpu hangs (for whatever reason)
|
| 09:39 | Bertl | so I was thinking, maybe we should have some 'poor men's debug interface' (i.e. basically a bunch of LEDs attached to one or more of the PMOD ports) to help in such situations?
|
| 09:43 | s3bastian | can you utilize the led array that is already on the zedboard?
|
| 09:43 | Bertl | yes, of course, but those are just 8 leds
|
| 09:44 | Bertl | i.e. even at optimal configuration only 256 states
|
| 09:44 | Bertl | I thought more about multiplexing 64 leds per PMOD port or so
|
| 09:47 | Bertl | or maybe even some kind of PC/usb interface, but for my purpose, I guess some leds would do
|
| 09:50 | Bertl | okay, off for now ... bbl
|
| 10:01 | dmj_nova | s3bastian: so are you suggesting the parallella as an additional board?
|
| 10:06 | dmj_nova | or are you specifically looking to add that epiphany coprocessor?
|
| 10:07 | s3bastian | tbh I don't know if it can be of any use for us, but maybe someone can think of a way
|
| 10:07 | s3bastian | Bertl, could the Zedboards Oled display be of any help fo debugging?
|
| 10:10 | dmj_nova | s3bastian: The epiphany coprocessor could be interesting
|
| 10:11 | dmj_nova | though the board itself is likely unable to interface quick enough to be a mainboard for us
|
| 10:13 | dmj_nova | s3bastian: I don't think we could actually get our raw data onto the board in realtime
|
| 10:22 | s3bastian | yes, the lack of GPIO is also one of my concerns
|
| 10:36 | dmj_nova | we can't really use the board as-is to process video in real-time
|
| 10:37 | dmj_nova | since ethernet is the fastest connection to the board
|
| 10:37 | dmj_nova | and that's 125MB/s
|
| 10:37 | dmj_nova | ...and we're likely to need ~180MB/s *average*
|
| 10:38 | dmj_nova | that said, the Epiphany IV chip might be useful
|
| 10:38 | dmj_nova | but I'm not sure
|
| 10:40 | s3bastian | do you know more about the epiphany chip?
|
| 10:40 | s3bastian | what exactly is it?
|
| 10:41 | s3bastian | btw I had to leve when we were in the middle of discussion hasing methods
|
| 10:42 | s3bastian | would it be a viable solution to hash the histogram to check image integrity instead of the entire image?
|
| 12:10 | Bertl | ad oled display: I thought about that, but it requires complicate SPI transfers, so not suitable for this purpose
|
| 12:11 | Bertl | ad hashing: I think we should definitely consider adding a crc check to the raw data, which in turn could be used for hashing (we already discussed that) but that doesn't help with the existing software framework which assumes full file skein hash and nothing else
|
| 12:13 | Bertl | ad debugging: I will come up with a simple solution, probably some kind of primitive PMOD board
|
| 12:14 | Bertl | dmj_nova: what about hdmi? we have at least 16bit at up to 200MHz IIRC
|
| 12:21 | Bertl | have to leave again, will be back in roughly two hours ...
|
| 13:14 | s3bastian | changed nick to: 77CAAJWGZ
|
| 13:14 | s3bastian | joined the channel |
| 13:14 | [1]s3bastian | joined the channel |
| 13:18 | [1]s3bastian | left the channel |
| 13:20 | dmj_nova | joined the channel |
| 13:24 | [1]s3bastian | joined the channel |
| 13:27 | s3bastian | left the channel |
| 13:27 | [1]s3bastian | changed nick to: s3bastian
|
| 13:34 | [1]s3bastian | joined the channel |
| 13:37 | s3bastian | left the channel |
| 13:37 | [1]s3bastian | changed nick to: s3bastian
|
| 13:44 | [1]s3bastian | joined the channel |
| 13:48 | s3bastian | left the channel |
| 13:48 | [1]s3bastian | changed nick to: s3bastian
|
| 13:54 | [1]s3bastian | joined the channel |
| 13:54 | [1]s3bastian | left the channel |
| 13:58 | s3bastian | left the channel |
| 14:14 | Bertl | back now ...
|
| 16:07 | Bertl | http://www.digilentinc.com/Data/Products/FMC-CE/FMC-CE%20Card%20User%20Guide.pdf
|
| 16:09 | Bertl | (interesting because of the complete FMC LPC and HPC pinout :)
|
| 18:15 | s3bastian | joined the channel |
| 18:40 | dmj_nova | s3bastian: it's a highly parallel coprocessor
|
| 18:42 | Bertl | hey, how's going?
|
| 18:43 | Bertl | http://www.adapteva.com/support/docs/e3-reference-manual/
|
| 18:43 | dmj_nova | Bertl: still visiting parents, going to a friend's grandma's funeral as well
|
| 18:43 | Bertl | (for the epiphany)
|
| 18:43 | Bertl | dmj_nova: oww, full program!
|
| 19:00 | s3bastian | I see
|
| 19:01 | s3bastian | but would it be of any use for us compared to custom logic blocks in the FPGA which are also as parallelized as we need them to be?
|
| 19:02 | Bertl | maybe, depends on what we want to do on the hardware
|
| 19:02 | Bertl | AFAICT, the main functions identified so far are (regarding AXIOM):
|
| 19:03 | Bertl | - high speed raw data acquisition from CMV12k
|
| 19:03 | Bertl | - various sensor data post processing
|
| 19:04 | Bertl | - HDMI/DVI output with overlay
|
| 19:04 | Bertl | - high speed data transport to recorder or
|
| 19:04 | Bertl | - high speed data recording directly to media
|
| 19:05 | Bertl | we also want to be able to configure all the sensor parameters, so probably we need some kind of GUI as well
|
| 19:05 | Bertl | i.e. touch screen input or similar
|
| 19:06 | Bertl | (please correct me or extend/explain where necessary)
|
| 19:11 | Bertl | on a different note: the FMC on the zedboard doesn't make me happy ... looking at the adapter schematics, I see that we are a few GPIO pins short and we have clock data on non clock lanes and there is one specifically ugly use of a reference voltage pin for exposure_time1
|
| 19:11 | Bertl | (while exposure_time2 is completely moved to the PMOD)
|
| 19:15 | Bertl | I'd propose to move all the 3.3V control blocks to PMOD, which would include the CLK_IN, SYS_RES_N, FRAME_REQ and the two exposure timer
|
| 19:16 | Bertl | and 'just' have all the LVDS data on the LPC-FMC
|
| 19:26 | s3bastian | sounds good
|
| 19:27 | s3bastian | yes those main functions are what I would list as well
|
| 19:31 | Bertl | okay good, I'll take a nap now if there isn't anything else :)
|
| 19:41 | s3bastian | please enjoy your nap :)
|
| 19:49 | se6astian | joined the channel |
| 19:51 | s3bastian | left the channel |
| 19:51 | se6astian | left the channel |
| 19:52 | s3bastian | joined the channel |
| 20:39 | se6astian | joined the channel |
| 20:41 | s3bastian | left the channel |
| 20:42 | s3bastian | joined the channel |
| 20:44 | se6astian | left the channel |
| 21:30 | se6astian | joined the channel |
| 21:32 | s3bastian | left the channel |
| 21:33 | s3bastian | joined the channel |
| 21:37 | se6astian | left the channel |
| 21:55 | s3bastian | left the channel |