Current Server Time: 05:14 (Central Europe)

#apertus IRC Channel Logs

2020/03/24

Timezone: UTC


00:28
Bertl_oO
off to bed now ... have a good one everyone!
00:28
Bertl_oO
changed nick to: Bertl_zZ
01:09
comradekingu
left the channel
02:42
JosephineSk
joined the channel
02:46
JosephineSk
left the channel
04:32
comradekingu
joined the channel
06:16
BAndiT1983|away
changed nick to: BAndiT1983
06:30
Bertl_zZ
changed nick to: Bertl
06:30
Bertl
morning folks!
06:30
lexano
left the channel
06:42
lexano
joined the channel
07:00
metal_dent[m]
morning Bertl !
08:29
se6ast1an
good morning
08:31
berto_bxl
joined the channel
08:57
bluez_[m]
Good morning everyone!
09:09
KarlieKloss
joined the channel
09:10
RexOrCine
joined the channel
09:12
berto_bxl
left the channel
09:13
KarlieKloss
left the channel
09:33
bluez_[m]
Bertl: Hey! So I did some research. I needed to ask a few questions...
09:33
Bertl
sure, go ahead
09:36
bluez_[m]
ok so... the driver I am going to make will use i2c to send commands to the pic? The pic, in response to these commands, will upload the "gateware" bitstream into macxo2's sram?
09:37
bluez_[m]
is this correct so far?
09:38
Bertl
yes
09:39
bluez_[m]
okay... so where is the gateware stored initially?
09:39
Bertl
on the SD card of the Axiom Beta
09:40
bluez_[m]
okay...so do we send that data too to the pic or the pic just responds to the command from main board and copies the gateware from sd card to the fpga?
09:41
Bertl
the PIC has no access to the SD card, so the data is sent over the I2C bus
09:42
bluez_[m]
ohh okay... so how do we retrieve the gateware from sdcard?
09:42
Bertl
for example with 'cat'
09:43
Bertl
as in 'cat /path/to/machxo2-gateware.bin >/dev/rfe'
09:43
bluez_[m]
i mean how will our driver do that? we need a script to first retrive that data?
09:43
Bertl
or alternatively the driver can request the firmware from a specific location
09:44
Bertl
(like it is done with other firmwares recently)
09:44
bluez_[m]
location in memory?
09:44
Bertl
filesystem
09:45
bluez_[m]
ohh okay
09:47
bluez_[m]
also if we are using i2c then why does the task mention "some kind of upload interface (JTAG, I2C, SPI)"
09:47
Bertl
that's the interface between PIC and MachXO2
09:48
Bertl
Zynq <-I2C-> PIC <-JTAG-> MachXO2
09:48
bluez_[m]
ok..how do i do that? the driver only works on the main board which runs linux right?
09:48
Bertl
correct
09:48
Bertl
the firmware for the PIC handles the translation
09:48
bluez_[m]
ohhh ok ok thats clear
09:49
bluez_[m]
so at first i only need to worry about sending commands to the pic along with gateware?
09:49
Bertl
yes
09:50
bluez_[m]
what about the second part "a debug interface (probably JTAG) to interface OpenOCD"
09:51
Bertl
what about it?
09:52
bluez_[m]
i mean... firstly, what does "interfacing OpenOCD" mean?
09:53
Bertl
you know what openocd is?
09:53
bluez_[m]
all i got is its for on chip debugging
09:53
bluez_[m]
just like jtag
09:53
Bertl
yeah, JTAG is the protocol/interface to do all kind of stuff including debugging
09:54
Bertl
OpenOCD is a software solution to allow access to such an interface via some kind of hardware
09:55
Bertl
for example, let's say you have a MachXO2 and you want to do some JTAG stuff on it
09:55
bluez_[m]
okay...
09:55
Bertl
you hook the jtag port up to some hardware dongle, let's say an FTDI chip
09:56
bluez_[m]
u mean a jtag programmer?
09:56
Bertl
then you configure OpenOCD properly and you can do things like retrieving data from the MachXO2
09:56
Bertl
yes, for example a JTAG programmer
09:58
Bertl
in our case, the JTAG programmer is the PIC
09:58
bluez_[m]
i see.... so openocd is a software sitting on the main board running under linux?... but how do we configure it to work "through" i2c to the pic...which than translates to jtag ?
09:58
bluez_[m]
<Bertl "in our case, the JTAG programmer"> ouuuu... okay makes sense
09:58
Bertl
well, that's part of the task to figure this out
09:59
Bertl
i.e. what is the best way to get the interface between OpenOCD and the PIC/JTAG working
09:59
bluez_[m]
is that related to making a driver??
10:00
Bertl
yes
10:00
BAndiT1983
Bertl: is it similar to LVP approach?
10:00
bluez_[m]
i see... where can i find more info about how the pic does stuff? Its seems a black box to me right now...
10:00
Bertl
BAndiT1983: not sure what you mean by that
10:00
BAndiT1983
datasheets
10:01
BAndiT1983
i mean if the signals for JTAG are generated similar to LVP ones, with toggling pins correctly
10:01
BAndiT1983
*-with +like
10:01
Bertl
yes
10:01
BAndiT1983
ok, interesting
10:02
Bertl
bluez_[m]: we use a PIC16F1718 and documentation can be found here:
10:02
Bertl
https://www.microchip.com/wwwproducts/en/PIC16F1718
10:03
Bertl
but unless you want to improve the firmware for the PIC, it is probably fine to consider it a black box device on the I2C bus which can do a few JTAG related tricks
10:03
bluez_[m]
<Bertl "https://www.microchip.com/wwwpro"> okay thank you!
10:05
bluez_[m]
Also, one final question, while writing the schedule, how detailed do I need to be? Cuz I am just learning right now I have an idea how it will go...but the details I can figure out while working?
10:05
BAndiT1983
weekly-based timeline is mandatory
10:05
BAndiT1983
to be able to verify the progress
10:06
bluez_[m]
<Bertl "but unless you want to improve t"> okay! Although, I needed to find out how the pic is programmed...i.e. what commands it will respond to and in what way...
10:07
Bertl
that is correct
10:08
bluez_[m]
<BAndiT1983 "to be able to verify the progres"> oh okay.... I'll share a draft after it is done then... can get your comments...
10:09
bluez_[m]
Bertl: Okay thanks a lot! It really helped. Working on the proposal then...
10:10
bluez_[m]
Oh and I more thing I forgot to ask ':D... the environment I need to setup... how am i going to test my code? remotely i guess but how exactly?
10:17
Bertl
we will provide remote access to an Axiom Beta, i.e. you logon via ssh and can work there
10:18
bluez_[m]
<Bertl "we will provide remote access to"> okay that works.. thanks again!
10:18
Bertl
you're welcome!
10:25
AaTYagi
joined the channel
10:28
RexOrCine
left the channel
10:45
AaTYagi
left the channel
12:59
berto_bxl
joined the channel
13:23
Bertl
off for now ... bbl
13:23
Bertl
changed nick to: Bertl_oO
13:36
RexOrCine
joined the channel
14:07
RexOrCine1
joined the channel
14:07
RexOrCine
left the channel
15:20
supragya_raj
joined the channel
15:39
RexOrCine1
left the channel
15:41
RexOrCine
joined the channel
15:44
supragya_raj
left the channel
16:02
berto_bxl
left the channel
16:09
Dest
joined the channel
16:09
Dest
hello
16:10
BAndiT1983
hi Dest
16:12
Dest
I've started on VHDL challenge and finished the separate UART & SPI-Master Modules
16:12
Dest
can someone have a look ?
16:12
Dest
https://drive.google.com/drive/folders/1Nl2zrLudWy8CafmEpvtH5W5cevLHMGmy?usp=sharing
16:13
BAndiT1983
good, but you would have to wait for Bertl_oO, as he is the mentor for VHDL
16:13
Dest
sure, thank you
16:27
BAndiT1983
changed nick to: BAndiT1983|away
16:30
berto_bxl
joined the channel
16:30
RexOrCine1
joined the channel
16:31
RexOrCine
left the channel
16:50
Dest
left the channel
17:17
BAndiT1983|away
changed nick to: BAndiT1983
17:44
BAndiT1983
changed nick to: BAndiT1983|away
17:51
berto_bxl
left the channel
18:13
CamilaMorro
joined the channel
18:17
CamilaMorro
left the channel
18:38
Shashwat
joined the channel
18:42
Bertl_oO
Dest: says I need permission to access ...
18:44
RexOrCine1
left the channel
18:45
berto_bxl
joined the channel
18:45
Bertl_oO
changed nick to: Bertl
18:51
berto_bxl
left the channel
19:16
Shashwat
left the channel
19:18
shashwat
joined the channel
19:26
Dest
joined the channel
19:26
Dest
oh sorry for that
19:27
Dest
would you check again now
19:30
Bertl
needs some formatting
19:31
Bertl
also it would be beneficial to have a short info how to build and test the project
19:34
Dest46
joined the channel
19:36
Dest
left the channel
19:37
Bertl
Dest46: you also might want to consider getting a proper IRC client instead of the web client
19:37
Dest
joined the channel
19:38
Dest46
left the channel
19:38
Dest
I will find another client
19:38
Dest
Regarding the formatting, can you elaborate a little more?
19:45
Bertl
for example proper spacing around expressions
19:46
Bertl
best take a look at existing code and try to fit in
19:46
Dest
left the channel
19:52
Dest123
joined the channel
19:54
Dest123
bretl: I've found a better client and I'll start working on your comments
19:54
Bertl
excellent!
20:21
shashwat
left the channel
20:26
BAndiT1983|away
changed nick to: BAndiT1983
20:27
Dest123
Should the SPI-Master send and receive from the UART at the same time?
20:44
Bertl
doesn't really matter
21:03
Dest123
This is how I imagine the bridge should be
21:03
Dest123
https://drive.google.com/a/zewailcity.edu.eg/file/d/1hFF7rYAhqarGxgekWEFm-yazuCAqdJ5s/view?usp=drivesdk
21:07
Bertl
best model some SPI slaves and test it then
21:48
preetimenghwani[
sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/JJQRkTWShEyrDEIPaJqervds >
21:50
Bertl
can't read long messages at the moment
21:51
preetimenghwani[
Oh I am really sorry
21:52
preetimenghwani[
I have some questions regarding T1130
21:52
Bertl
let's hear then ..
21:53
preetimenghwani[
I understood pixel remapping but how can that help in compaction of data stream
21:54
preetimenghwani[
Also how can rearrangement and repacking help in compaction
21:55
Bertl
the DDR memory is accessed via AXI bus interface
21:55
Bertl
and to achieve maximum throughput, burst writes are required
21:56
Bertl
this means that blocks of 64/128 bit need to be sent
21:57
Bertl
let's say the sensor is operating in 12 bit mode, then the sensel data needs to be rearranged to fit into those 64/128 bit words
21:58
Bertl
this is what the task description refers to as (re)packing
21:59
Bertl
the sensel data also does not come as one sensel at a time ... depending on the sensor configuration up to 64 sensel are shifted in in parallel
22:00
Bertl
as additional complication, those pixel are spread out over one or two lines and thus require rearrangement to be stored in order
22:04
preetimenghwani[
By 12 bit mode you mean that pixels store 12 bit data right?
22:05
Bertl
yes
22:05
intracube
left the channel
22:06
preetimenghwani[
So the packets will be of size 128*12 and that we need to break into 64/128 bits just confirming
22:07
intracube
joined the channel
22:07
Bertl
the sensor has a number of data outputs (up to 64 in the CMV12k case)
22:08
Bertl
and each data output transmits a sensel (8, 10 or 12 bit) serially
22:09
Bertl
the sensel are from spaced colums, e.g. 0, 128, 256, ...
22:10
Bertl
the data is deserialized in parallel and needs to be rearranged into a sequence of 64/128 bit words for storing into DDR memory via the AXI bus
22:12
Bertl
e.g. for 12bit, and 64bit words, 3 words i.e. a total of 192bit would work
22:16
preetimenghwani[
You said each data output transmits a sensel but LVDS outputs transmits 128 pixels at a time
22:17
preetimenghwani[
I am sorry for bothering you with silly questions
22:17
Bertl
did I say that?
22:18
Bertl
at most 64 sensel are transferred in parallel (for the CMV12k)
22:18
preetimenghwani[
Oh okay my bad
22:20
Bertl
and no problem regarding questions, that's what we are here for at the moment
22:27
preetimenghwani[
The task says we need to implement remapping strategies does it mean that we need to modify existing ones with lower gate count?
22:29
Bertl
there is an existing remapper but it is very limited
22:29
Bertl
and yes, the goal is to optimize gate count and performance
22:31
preetimenghwani[
Very limited in what sense?
22:32
Bertl
it only handles 12bit and 32 sensel at the moment
22:33
Dest123
hey bretl, I was wondering how are challenging tasks going to be evaluated
22:34
Bertl
depends on the tasks, but in general we review them, provide some feedback, see how the student improves/adapts and draw our conclusions
22:38
Dest123
Should tasks supporting additional functionality have a higher acceptance rate?
22:40
Bertl
if the functionality makes sense in the context and doesn't bloat the implementation unnecessarily, then yes
22:41
preetimenghwani[
There are remapping strategies for binning and subsampling we need to improvise the strategiea for both?
22:41
Bertl
Dest123: note that attention to detail is better than producing tons of code
22:42
Bertl
the sensor supports a number of 'modes' which affect how the data is transmitted
22:42
Bertl
i.e. the semantics of the data, not the actual transmission
22:43
Bertl
switching the mode requires the remapper to accomodate for this, otherwise the data will be out of sequence
22:44
preetimenghwani[
By modes you mean 162 or 321 right?
22:45
preetimenghwani[
Sorry 32 x 1 or 16x 2
22:45
Bertl
more the subsampling, binning, windowing and number of output channels
22:52
preetimenghwani[
Thanks a lot Bertl that was really helpful!
22:52
Bertl
you're welcome!
22:56
BAndiT1983
changed nick to: BAndiT1983|away
23:37
Dest123
left the channel
23:58
Spirit532
left the channel
23:59
Spirit532
joined the channel