**** BEGIN LOGGING AT Wed Mar 29 03:00:04 2017
Mar 29 07:29:15 <ravikp7> jkridner: I cross compiled u-boot with default ums mode, here are the binaries https://goo.gl/1oissp for testing
Mar 29 10:27:17 <gcl-bot> [abhishek-kakkar] jkridner: Can you send me a mentor invite link?
Mar 29 11:29:12 <jkridner> Abhishek_: k
Mar 29 12:17:34 <kiran4399> jkridner: I created something called beagleCV
Mar 29 12:17:52 <kiran4399> from a libCVD repository
Mar 29 12:19:01 <kiran4399> jkridner: I want to develop Computer vision support for beaglebone and create support for the SGX530.. I will be doing the stereo vision thing as an example to illustrate the use of BeagleCV.
Mar 29 12:19:05 <kiran4399> what do you think?
Mar 29 12:19:21 <kiran4399> jkridner: I already talked to ds2 about this.. what do you think?
Mar 29 12:19:25 <kiran4399> m_w: ^^
Mar 29 12:19:31 <kiran4399> nerdboy: ^^
Mar 29 14:27:43 <jkridner> kiran4399: if possible, very cool.
Mar 29 14:31:49 <kiran4399> jkridner: can you please look at it once? http://elinux.org/BeagleBoard/bb_blue_api_v2
Mar 29 14:33:08 <kiran4399> jkridner: please don't mind the formatting errors... there is still lot to edit.. let me know about the idea..
Mar 29 14:57:47 <jkridner> kiran4399: any other proposed mentors?
Mar 29 15:20:15 <jkridner> kiran4399: is there previous art on using OpenGLES2 for stereo vision?
Mar 29 15:33:21 <jkridner> http://queue.acm.org/detail.cfm?id=2206309 is too long to read
Mar 29 15:34:30 <jkridner> http://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=7761572&fileOId=7761597 also looks long
Mar 29 15:34:43 <AndChat|361524> jkridner: there is no much support in opengl es2
Mar 29 15:34:53 <jkridner> right, that's my fear.
Mar 29 15:35:21 <AndChat|361524> jkridner: why?
Mar 29 15:36:40 <jkridner> http://jultika.oulu.fi/files/isbn9789526206721.pdf seems to have some analysis
Mar 29 15:36:51 <jkridner> AndChat|361524: because kiran4399 is looking to use it in his proposal.
Mar 29 15:38:17 <jkridner> http://www.ee.oulu.fi/~miguelbl/publications/Paper%20CVPR12.pdf is also interesting
Mar 29 15:38:29 <jkridner> seems to be same research
Mar 29 15:46:25 <jkridner> https://pdfs.semanticscholar.org/fafc/75bea7a94e12bc315f9b4c4d6b3bc397ad8e.pdf also looks interesting, but nothing clear without a long read there either.
Mar 29 15:48:08 <jkridner> http://jcse.kiise.org/files/V10N2-04.pdf  <--- more stuff to read
Mar 29 15:53:17 <kiran4399> jkridner: I am sorry.. I am AndChat|361524
Mar 29 15:53:52 <kiran4399> jkridner: I forgot to mention it.. there is no much support in opengl es2 for computer vision stuff.. but there is support for graphics and all..
Mar 29 15:58:34 <kiran4399> jkridner: what do you think of this project?? Any suggestions/
Mar 29 15:58:58 <jkridner> figuring out this stuff ahead of time (ie., is it really doable) is my suggestion.
Mar 29 15:58:59 <kiran4399> ds2: http://elinux.org/BeagleBoard/bb_blue_api_v2
Mar 29 15:59:24 <jkridner> I love the idea of doing some stereo vision and object recognition acceleration support.
Mar 29 16:02:09 <kiran4399> jkridner: btw.. ds2 is very much interested in this..
Mar 29 16:03:54 <jkridner> kiran4399: that's really great to hear.
Mar 29 16:04:16 <kiran4399> jkridner: this link is helpful.. http://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=7761572&fileOId=7761597 thanks..
Mar 29 16:05:34 <kiran4399> jkridner: The paper's approach is almost similar to mine..
Mar 29 16:21:13 <m_w> pmezydlo: I posted the start of the fpga schematic
Mar 29 16:21:29 <kiran4399> m_w: will you mentor me on this?
Mar 29 16:21:37 <kiran4399> m_w: http://elinux.org/BeagleBoard/bb_blue_api_v2
Mar 29 16:21:47 <pmezydlo> great,
Mar 29 16:22:47 <m_w> kiran4399: possibly
Mar 29 16:23:00 <m_w> I have quite a few potential students
Mar 29 16:23:12 <kiran4399> m_w: can you once review it?
Mar 29 16:23:26 <m_w> kiran4399: sure
Mar 29 16:23:42 <pmezydlo> m_w: I see repository :)
Mar 29 16:24:14 <m_w> pmezydlo: fork the repository to your github, and then add your name to the license and do a pull request
Mar 29 16:24:51 <m_w> then try to find a good regulator for the 3.3V and 1.2V rail
Mar 29 16:25:12 <pmezydlo> yeah of course I will do it now
Mar 29 16:29:17 <pmezydlo> I will find, texas instruments offer good regulators
Mar 29 16:29:57 <ravikp7> jkridner: did you test the binaries ?
Mar 29 16:29:58 <thetransformerr> hi everybody
Mar 29 16:30:38 <thetransformerr> nerdboy: as ds2: and you asked yesterday, I analysed the required sampling rate
Mar 29 16:31:19 <thetransformerr> I found PRUDAQ has way too much sufficient sampling capacity
Mar 29 16:31:47 <thetransformerr> but drawbacks are it has only two channels
Mar 29 16:32:42 <m_w> kiran4399: bb_blue_api_v2 do not seem to capture the idea of the project
Mar 29 16:32:48 <thetransformerr> nerdboy: so when planning for three axis, I think we would be having some complexity, what do you suggest
Mar 29 16:33:19 <kiran4399> m_w: I am sorry.. I made that change long ago.. I think I will change the website.. give me a minute..
Mar 29 16:33:44 <m_w> Stereo Vision support for BeagleBone using BeagleCV <= so this the real project name?
Mar 29 16:33:50 <jkridner> aren't we 30 minutes into "meeting time"?
Mar 29 16:34:05 <ds2> hehehe
Mar 29 16:34:11 <m_w> meetings started already?
Mar 29 16:34:59 <kiran4399> m_w: yeah..
Mar 29 16:35:08 <kiran4399> m_w: yeah..
Mar 29 16:35:17 <m_w> okay calling to order :)
Mar 29 16:36:07 <ds2> chaos++;
Mar 29 16:36:39 <kiran4399> ds2: http://elinux.org/BeagleBoard/beaglecv_stereo
Mar 29 16:36:47 <Mala> Hello beagleboard!
Mar 29 16:36:51 <kiran4399> m_w: http://elinux.org/BeagleBoard/beaglecv_stereo
Mar 29 16:38:21 <thetransformerr> oh sorry.... PRUDAQ actually has 2 channels but it also has got 4x1 switch , so in effect it can have 8 input channels
Mar 29 16:38:38 <ds2> who's who?
Mar 29 16:40:32 <ds2> student candidates vs ???
Mar 29 16:40:51 <Mala> vs beagle admin
Mar 29 16:41:02 <kiran4399> ds2: what is your user page on elinux?
Mar 29 16:41:07 <kiran4399> m_w: ^^ same
Mar 29 16:41:22 <jkridner> I'm a BeagleBoard.org GSoC admin. :-)
Mar 29 16:41:33 <Mala> Jkridner!
Mar 29 16:41:45 <jkridner> my elinux page is http://elinux.org/User:Jkridner
Mar 29 16:41:52 <Mala> There's nerdboy too
Mar 29 16:42:02 <Mala> That's 2 admin here at least. How are you jk?
Mar 29 16:42:54 <Mala> Ds2, according to the site, m_w and one of the av500 are also admin for beagleboard.
Mar 29 16:44:03 <ds2> I am hyau2 on elinux
Mar 29 16:45:34 <ravikp7> Mala: you're probably confused between mentors and admin
Mar 29 16:45:36 <thetransformerr> so if everything is sorted, lets start over again
Mar 29 16:45:43 <ds2> not many students here and isn't this the last week of open app?
Mar 29 16:45:48 <jkridner> Mala: please try to avoid direct messages to me unless really private. questions about projects shouldn't be private.
Mar 29 16:45:55 <Mala> Ok
Mar 29 16:46:13 <jkridner> Mala: lots are very new to this process, so don't be scared of sounding silly.
Mar 29 16:46:20 <jkridner> we are a bit "trial by fire" here.
Mar 29 16:46:42 <Mala> Are we submitting proposals that Beagleboard wants, or things we want that we think Beagleboard lines up with? Or is Beagleboard actually just oversight?
Mar 29 16:46:52 <jkridner> the expectation is to fight through the desire to work in isolation and move to working wide-open.
Mar 29 16:47:09 <ds2> kiran4399: will read through it in a bit. Is the first URL obsolete now?
Mar 29 16:47:21 <jkridner> proposals should align with where you can get BeagleBoard.org mentors to support you.
Mar 29 16:47:34 <kiran4399> ds2: yeah..
Mar 29 16:47:42 <jkridner> the mentors have posted some ideas they have on projects that would be useful to the community and interesting to them to support.
Mar 29 16:48:01 <jkridner> proposals with well-engaged mentors will be the most likely to be selected.
Mar 29 16:48:10 <jkridner> fair statement m_w, nerdboy, av500?
Mar 29 16:48:13 <m_w> kiran4399: http://elinux.org/User:M_w
Mar 29 16:49:23 <jkridner> a topic that has come up much lately is that most people don't know the am3358 has a Cortex-M3 core.
Mar 29 16:49:36 <Mala> Potential silly question, where can we find the ideas they posted?
Mar 29 16:49:46 <jkridner> http://bbb.io/gsocideas
Mar 29 16:50:00 <jkridner> everything should be linked off http://bbb.io/gsoc
Mar 29 16:50:16 <ordsen> m_w: apropos proposal: guess I have to bother you again with my 2nd prop
Mar 29 16:51:10 <jkridner> Mala: how did you find the chat if you didn't see the ideas page?
Mar 29 16:51:33 <Mala> jkridner: I found the ideas page. Thanks!
Mar 29 16:51:34 <ds2> the am335x series is awash in different cores
Mar 29 16:52:02 <m_w> ordsen: link?
Mar 29 16:52:08 <ordsen> m_w: as we discussed yesterday I intended to focus in my 2nd proposal more on the stretch goals of my 1st proposal which target serial control and web API. But I just saw that the other applicant for the terminal server now also included the serial control features; I think that my 2nd proposal will replicate too much functionality of his proposal..
Mar 29 16:52:57 <ordsen> still have the draft on my pc. when everything is clear I will overhaul it and upload it
Mar 29 16:53:09 <m_w> ordsen: yeah looks like it is going to be just one or the other of you
Mar 29 16:53:40 <ordsen> m_w: I’m considering to leave out all the serial communication stuff and add Bluetooth and GPIO control through a generic interface.
Mar 29 16:54:23 <m_w> doesn't the nodered or other preinstalled web stuff on the bone already do some of that>
Mar 29 16:54:26 <m_w> ?
Mar 29 16:55:26 <ordsen> it can be extended to do that as I see it.
Mar 29 16:55:26 <m_w> I think you need to differentiate from those to really impact the community
Mar 29 16:56:37 <ds2> kiran: sgx isn't it?
Mar 29 16:58:13 <ordsen> m_w: What if "installing" new devices (e.g. capes) into the API can be done easily/dynamically and addressing is done in the same way other device (which are connected to the API) are addressed?
Mar 29 16:59:30 <ordsen> to achieve this one has to put much effort into the existing web stuff. Having a framework for that sounds reasonable to me
Mar 29 16:59:54 <ds2> kiran4399: seems like a lot... prehaps drop the camera integration?
Mar 29 17:00:17 <kiran4399> ds2: Alright..
Mar 29 17:00:25 <kiran4399> ds2: Even I felt the same..
Mar 29 17:00:35 <kiran4399> ds2: I will drop it.. Anything else?
Mar 29 17:02:10 <ds2> more details
Mar 29 17:02:22 <ds2> there is a lot of fluff
Mar 29 17:02:27 <thetransformerr> nerdboy: I read about PRUDAQ and found that it has two channels, where each channel have a 4x1 analog switch for inputs, hence
Mar 29 17:02:59 <m_w> ordsen: sounds good, ellaborate on the proposal
Mar 29 17:03:23 <ds2> thetransformerr: what sample rate do you need?
Mar 29 17:03:41 <kiran4399> ds2: means?
Mar 29 17:03:43 <thetransformerr> it can take up 8 input but it becomes complex due to the fact that sample result arrive after 3 cycles of 25 ns
Mar 29 17:03:55 <kiran4399> ds2: is it too much technical stuff?
Mar 29 17:04:29 <ds2> kiran4399: no no... lots of intro stuff but not much details.... it is nice that it can go into the blue api eventually but that doesn't tell me more about this
Mar 29 17:05:01 <ds2> kiran4399: it can be used for more then just that... offloading the A8 can have many benefits...right now, the SGX is barely used in embedded stuff
Mar 29 17:05:22 <thetransformerr> ds2: for a 40khz transducer we would be requiring a sample rate of 400,000 per second
Mar 29 17:05:44 <thetransformerr> and that is easily available with PRUDAQ
Mar 29 17:05:54 <ordsen> m_w: so basically, it can be seen as a bus accessible through the network
Mar 29 17:05:59 <ds2> thetransformerr: so 2 channels of that?
Mar 29 17:06:47 <thetransformerr> yup, ds2: as I said it is going to be cakewalk with PRUDAQ for 2 axis sonic anemometer
Mar 29 17:06:55 <ordsen> m_w: devices will have to present themselves to the bus with some kind of config file (possibly some json/xml)
Mar 29 17:07:33 <thetransformerr> but 3 axis sonic is going to be complex due to the fact that
Mar 29 17:07:37 <ds2> maybe
Mar 29 17:08:03 <ordsen> m_w: some kind of broadcasting observers of events issued by a device would be nice
Mar 29 17:08:28 <thetransformerr> This is somewhat complicated if we want to round-robin between inputs on every clock edge, since the ADC has a 3-cycle pipeline latency -- the data lines show you what was captured 3 cycles ago, while the select lines show you what's being captured now.
Mar 29 17:09:04 <ds2> that almost sounds like a project in itself
Mar 29 17:09:37 <thetransformerr> ds2: that almost sounds like a project in itself ?? is it for me
Mar 29 17:09:50 <ds2> thetransformerr: yes
Mar 29 17:10:32 <kiran4399> ds2: so where is it that you want me to describe?? GPU part or beaglecv part or stereo part?
Mar 29 17:10:50 <thetransformerr> all these limitations we are going to face in, when in future if we would require to scale out our project for other capablities
Mar 29 17:11:58 <ds2> kiran4399: all of it... you are spending a lot of time stripping out libcvd...the GPU is a black box... prehaps explain a little on how GLES can be used to compute (the usual stuff about shaders, etc) along with datatype mapping
Mar 29 17:12:53 <thetransformerr> still I say PRUDAQ may be best DAQ considering its cost of $79, because all other DAQ I saw today from companies like National Instruments were way too much costly like $400-500
Mar 29 17:12:54 <kiran4399> ds2: Oh.. yeah.. go tit..
Mar 29 17:13:24 <ds2> kiran4399: I am not the only reader of it
Mar 29 17:13:33 <kiran4399> ds2: btw.. I forgot to mention.. I just ran libcvd and did a fast corner detection test..
Mar 29 17:13:51 <kiran4399> ds2: it ran perfrectly.
Mar 29 17:14:20 <thetransformerr> ds2: What would you suggest about DAQ
Mar 29 17:14:31 <thetransformerr> any thoughts
Mar 29 17:15:14 <Mala> jkridner: At this point in bonescript are you still aiming to add features, or mostly deal with bug queues?
Mar 29 17:15:37 <thetransformerr> nerdboy: we can consider transducers of high frequency also like of 120khz???
Mar 29 17:15:40 <ds2> thetransformerr: donno of ideas for 3 channels...
Mar 29 17:16:29 <ds2> at 400KHz, I wonder if you can use another bus...the PRUDAQ stuff depends on PRU firmware and you can get stuck/spend a lot of time on that part w/o working on the real project :(
Mar 29 17:16:50 <ds2> kiran4399: saw your message on that... thought you said it had some GLES stuff in it?
Mar 29 17:21:01 <thetransformerr> yeah actually 400khz is the best of best required sampling rate, and although I haven't completed my analysis, about this part because this was scheduled for later,
Mar 29 17:21:30 <thetransformerr> ds2: but there are are many projects that I think have achieved it
Mar 29 17:22:33 <thetransformerr> and ds2: as the project idea suggest the most important part of this Sonic anemometer is this sampling and PRU stuff only
Mar 29 17:23:42 <thetransformerr> also there is a BeagleLogic software that is compatible with PRUDAQ AND OFFERS A MUCH HIGH sampling capacity
Mar 29 17:24:52 <nerdboy> oops
Mar 29 17:25:06 <thetransformerr> and ds2: "at 400KHz, I wonder if you can use another bus...the PRUDAQ stuff depends on PRU firmware and you can get stuck/spend a lot of time on that part w/o working on the real project :(" I think we had a misunderstanding
Mar 29 17:25:21 <ds2> verify exactly what works.... IIRC - there are limitations the last time I spoke to Absheik
Mar 29 17:25:55 <thetransformerr> 400kHz is rate required for transducer output sample
Mar 29 17:25:57 <ds2> BeagleLogic is async capture
Mar 29 17:26:21 <ds2> so either you have to post process to get samples out or you need to make it synchronous
Mar 29 17:26:33 <ds2> thetransformerr: what sample rate do you need from the DAC then?
Mar 29 17:27:54 <nerdboy> you're supposed to be analyzing thode reqs...
Mar 29 17:28:02 <nerdboy> *those even
Mar 29 17:29:23 <nerdboy> you can spec the sensors (hopefully within the hobby-ish price range) based on what you need for prudaq and approx. distance between the sensors
Mar 29 17:30:08 <thetransformerr> yeah nerdboy: and ds2:, I am doing detailed analysis this time since yesterday and I am in middle of it,
Mar 29 17:30:12 <nerdboy> there should be optimal frequency range for getting decent signals
Mar 29 17:30:41 <nerdboy> the muratas are one example
Mar 29 17:30:43 <thetransformerr> I was here just to get suggestions here
Mar 29 17:31:04 <nerdboy> they are separate tx and rx parts
Mar 29 17:31:29 <thetransformerr> yeah I have read spec sheet of several others like microcomp, sonix etc
Mar 29 17:31:58 <nerdboy> make a table
Mar 29 17:32:27 <kiran4399> ds2: yeah.. it has some helper functions..
Mar 29 17:32:50 <kiran4399> ds2: https://codedocs.xyz/edrosten/libcvd/group__gGL.html
Mar 29 17:33:14 <thetransformerr> "they are separate tx and rx parts" we require here a transducer, I am sure about it else equations are going to mess up with separate tx and rx
Mar 29 17:33:36 <thetransformerr> transreceiver
Mar 29 17:36:13 <ds2> kiran4399: a discussion of how that matches up to GLES could be useful... that URL/API looks like plain GL support
Mar 29 17:37:06 <pmezydlo> m_w: tomorrow I have exam from verilog :) I have to prepare myself.
Mar 29 17:37:08 <pmezydlo> I did pull request. Tomorrow I will to prepare power management proposal to beaglewire.
Mar 29 17:38:56 <pmezydlo> m_w: I have to add all required kicad library because for now I have warning messages
Mar 29 17:39:02 <nerdboy> it should work either way, some are less easy to control then others
Mar 29 17:46:50 <thetransformerr> ds2: "BeagleLogic is async capture" "[22:56] <ds2> so either you have to post process to get samples out or you need to make it synchronous" would you explaining it please because as per blog written by abhishek
Mar 29 17:47:19 <kiran4399> ds2: well.. it written on the readme that opengl es support is there.. and moreover.. opengl does not work on the bone right?
Mar 29 17:47:23 <kiran4399> so it must be opengl es.
Mar 29 17:47:32 <kiran4399> ds2: right?
Mar 29 17:48:16 <thetransformerr> he says "The dual-PRU architecture of BeagleLogic that maintains a clear separation between the sampling operation itself (done by PRU1) and data transfer to memory (done by PRU0) means that on an application level, only the firmware on PRU1 needs to be changed in order to support PRUDAQ, no modification is necessary on the kernel driver side, raw analog data can be directly read through /dev/beaglelogic and this also o
Mar 29 17:48:41 <m_w> pmezydlo: okay good luck with your test, I will fix the warnings
Mar 29 17:49:20 <pmezydlo> m_w: I don't have latice lib
Mar 29 17:50:12 <pmezydlo> can i download it from your kicad repository?
Mar 29 17:51:06 <m_w> it was a library from xess
Mar 29 17:51:20 <m_w> I can embedded the lib into the project to make it self contained
Mar 29 17:52:09 <m_w> if you want to work on anything after your exam you can update the README with some useful information
Mar 29 17:59:17 <uka_in> m_w,nerdboy: i tried to enable ums mode.. but was stuck with an error https://pastebin.com/hsUsFvcF ,please have a look
Mar 29 18:01:16 <uka_in> also, since FSG_NUM_BUFFERS=2 by default, is this value ok with am335xx?
Mar 29 18:01:30 <m_w> https://lists.denx.de/pipermail/u-boot/2016-April/252542.html
Mar 29 18:35:52 <uka_in> m_w: i did "make config" and enabled ums..still having that error
Mar 29 18:36:33 <uka_in> i added following to the configurations https://pastebin.com/1TtR6DK9
Mar 29 18:36:43 <uka_in> did i missed something?
Mar 29 18:39:10 <m_w> uka_in: grep for the function names in the source code
Mar 29 18:39:31 <m_w> find if they have been implemented for the USB on the beagle
Mar 29 18:40:03 <m_w> find the configuration option that adds that file in the associated Makefile
Mar 29 18:40:38 <uka_in> ok
Mar 29 18:41:30 <m_w> so it looks like I may be starting a group buy for the pocketbone if anyone is interested
Mar 29 18:42:04 <m_w> the trick is getting enough people to make the price reasonable
Mar 29 18:51:36 <m_w> https://hackaday.io/project/19495-pocketbone-kicad
Mar 29 18:57:24 <nerdboy> i'm in unless they're way overpriced...
Mar 29 19:00:06 <m_w> if we get 26 orders it will be ~$100 a piece
Mar 29 19:00:44 <m_w> if we get 100 orders it will be ~$66 a piece
Mar 29 19:02:26 <m_w> the most I can put on my credit card is 36 and that would be ~90 a piece
Mar 29 19:21:29 <nerdboy> i think a beaglebone condig needs to be teased out of the em335_evm stuff
Mar 29 19:21:40 <nerdboy> *in u-boot
Mar 29 19:21:57 * nerdboy looking at mainline/master branch
Mar 29 19:22:20 <nerdboy> the storage module isn't being enabled
Mar 29 19:23:21 <nerdboy> i just hacked include/configs/am335x_evm.h to make it build but i have no idea if it works...
Mar 29 19:44:10 <nerdboy> http://sprunge.us/UZdB  <= feel free to test it
**** ENDING LOGGING AT Thu Mar 30 03:00:01 2017