#cubox irc log of Tue 25 Aug 2015

IRC log of #cubox of Tue 25 Aug 2015. All times are in CEST < Back to index

10:09 topi`_ rabeeh: do you offer a Extended temp variant of the HB Edge? At least one component is critical - the eMMC (we have first hand experience of MLC flash corruption when doing writes at sub-zero temperatures and the component being of the commercial variant)

10:10 topi`_ if the Edge does not offer such a variant, then maybe we would need a custom batch of boards with such an eMMC in place?

11:05 rabeeh topi`_: yes there are extended variants too

11:06 rabeeh the dual and quad devices are only extended and beyond; we don't use commercial versions of them

11:06 topi`_ right.

11:06 rabeeh the eMMC device by default is a Samsung 4GByte device.

11:06 topi`_ I'm not worried about the i.MX6, I'm worried about the eMMC

11:06 topi`_ :)

11:06 topi`_ even if the processor fails to start when it's too cold, that's not destructive, however, if the eMMC gets destroyed then that means a field trip for our ops

11:07 topi`_ a few months back we checked the different MMC vendors, and the extended temp variants are usually significantly more expensive than the commercial ones

11:09 rabeeh KLM4G1FEPD

11:10 rabeeh ^^^ This is the part we are using.

11:18 topi`_ rabeeh: got it. Weird thing is, the codes about that part have nothing that describes the temperature grades for that chip...

11:21 topi`_ actually it might be that all Samsung eMMC parts are for T -25 .. 85

11:21 topi`_ I found one Samsung eMMC with very similar code that is -25..85

11:21 rabeeh i have the spec somewhere.

11:24 topi`_ that chip seems to be quite new, sampling Q1'15

11:24 topi`_ maybe it works with eMMC 5.0, however I guess the i.MX6 SD controller won't talk MMC5.0?

11:35 topi`_ on Samsung's eMMC site, I see only -25..85 C devices and the -40..105 devices that are labeled "automotive"

11:35 topi`_ so maybe that eMMC is just fine for our purposes

11:37 jnettlet topi`_, find yourself an industrial freezer and get to testing :)

11:41 mk01_ .... I can provide testing to -74C (if would be really needed)

11:41 jnettlet mk01_, are you going on an Artic mission?

11:42 topi`_ jnettlet: here in Finland, we just wait for the winter to come, and then can easily test at below -25C :)

11:43 jnettlet sure, bottle of vodka, long cable, and a sauna. You can test both extremes and enjoy yourself doing it :)

11:44 mk01_ jnettlet: no, but neighbour's daughter married a penguin

11:44 mk01_ ;)

11:44 jnettlet Tux?

11:44 jnettlet :_

11:44 mk01_ pengu

11:44 mk01_ yup

11:46 mk01_ ((-74C is paper wise needed to store bio samples over 4h. so it's not that uncommon to find (those freezers))

11:47 topi`_ just pour a bottle of liquid CO2 and you have -78C :)

11:47 mk01_ yes, allowed. up to 4h

11:47 mk01_ as said :))

11:48 jnettlet dry ice works as well

11:48 topi`_ installing python deps via pip on i.MX6Q ... it takes quite a lot of time :/

11:49 topi`_ unfortunately none of these seem to be using multiple cores

11:49 topi`_ not even the compile phase; it just launches cc sequentially

11:52 mk01_ jnettlet: have you changed in any way the g2d part in videocodecimx ?

11:52 topi`_ rabeeh: are there any i2eX microSOM's that would offer 2GB of DRAM? it seems the imx6 Dual is only available with 1GB ram?

11:52 jnettlet mk01_, yes

11:53 mk01_ a easy ones or more rewrite ?

11:53 mk01_ a = an

11:53 jnettlet a bit of both.

11:53 jnettlet I will post them when I am doine

12:00 mk01_ because I finished rewriting the general part of CIMXContext (it doesn't need stop/start anymore and fb reconf runs on the fly). I would send that as 2nd PR to upstream, but if you rewrote that other part, I will wait and sync with your result...

12:01 mk01_ makes sense, or ?

12:01 jnettlet no go ahead and I will rebase on your patches. I am pretty overloaded right now.

12:01 mk01_ ok, no pushing

12:01 jnettlet Although I think I have fixed the bottleneck using the VPU -> IPU direct rendering

12:05 mk01_ ok, so you removed the circle around *kodi (to get the data from line 1 to line 2?) ?

12:06 mk01_ do you have any draft somewhere, to let me see what will still be valid from the code ?

12:35 mk01_ jnettlet: did you give any thoughts to the PAL/NTSC setup question ? I was going through the logs on from #openelec and you was asking if I PR-ed the fractional modes - basically the commit on the side of xbmc would be 4 lines - so nothing to send in "advance"

12:42 jnettlet mk01_, lets go with what you have for now. If we decide we need to tweak it we can work from that.

12:42 mk01_ ok

12:42 jnettlet like I said before, this will be revisited when we switch to KMS so I don't want to dump a bunch of time into it now.

12:43 mk01_ then in the g2d section I just fixed the align on "zooming" out of screen

12:43 jnettlet are you scaling up and and then cropping?

12:43 mk01_ I cpuldn't find any other misalignments (tested 1600x900 too, 1440x900 and 1366x768

12:44 jnettlet I actually found one yesterday. Very strange. It was at 1080p but the video was low quality, like DVD resolution.

12:44 mk01_ jnettlet: yes, that one

12:44 mk01_ ^^^^ to answer your scale + scrop

12:45 jnettlet there the entire picture had that wrapping even without the overlay being shown

12:45 jnettlet okay, if you post your pull request I will pull it down and test later.

12:46 mk01_ jnettlet: what means: "had that wrapping ... " ?

12:47 mk01_ ah maybe I know what you mean

12:47 jnettlet where the bottom of the picture was wrapping around the framebuffer and being shown at the top. It was flickering so I assumed the pageflipping was wrong

12:47 mk01_ yep

12:47 mk01_ ok

12:47 mk01_ this is what I was telling you before. just coould explain.

12:47 jnettlet was strange it showed up on normal video playback

12:47 jnettlet okay

12:48 jnettlet is this all pushed to your kodi on github?

12:48 mk01_ nope, just cleaning the code.

12:49 mk01_ 20m

12:49 mk01_ ok what I want to say -

12:49 mk01_ I can see that too. but only IF doublerate deint is on

12:50 mk01_ and the content has actually no top/bottom info

12:50 jnettlet I am pretty sure this was progressive, but it is possible it was interlaced.

12:50 mk01_ so all is NONE

12:50 jnettlet just some random internet video

12:51 mk01_ ok, to be even more clear - content with flag Interlaced, but no frames other from VPU_FIELD_NONE

12:51 mk01_ + active deint and choosed doublerate

12:52 mk01_ this is when It happens always (for me)

12:53 jnettlet fun

12:54 mk01_ can you also test some .iso images of DVDs with VPU decoding ?

12:55 mk01_ by default it is forced to ffmpeg, but I have great success with VPU decode. just mine 'test' pool of such media is TWO

12:55 mk01_ so great success = 100% of two files

12:56 jnettlet mk01_, okay that should work if decss can keep up.

12:56 jnettlet or are these already stripped?

12:57 mk01_ mmnt

13:05 mk01_ ok, not sure I 'm able to give a proper answer

13:05 mk01_ but the content seems to be properly unpacked , yes

13:05 mk01_ ((that's why I'm asking if you can check))

13:08 jnettlet sure, my wife wants me to rip a bunch of her old DVD's anyways

13:59 mk01_ jnettlet: here it is - https://github.com/xbianonpi/xbian-sources-xbmc/commit/ff9f265c146cddb34bbf22cdff08c5c8d9ba33e7 - but it is absolutely unreadable even as WIP. so have to split it into at least two hunks. one with actual code changes, the other the rest (mostly delete + reorder). later today. ok?

14:00 jnettlet mk01_, sure no rush. Like I said I am pretty slammed. Just trying to make sure we keep going in a forward direction :)

14:09 mk01_ jnettlet: here is the ffmpeg force so you don't have to look for it https://github.com/xbianonpi/xbian-sources-xbmc/commit/707beaa356efafa03f1de9aaa1c6d4f9e29bc8da

14:09 mk01_ (DVDs)

14:09 jnettlet mk01_, sure which program are you using to image the DVD's?

14:11 mk01_ that's the thing I explained - I have just two DVDs. otherwise all from iTunes.

14:12 mk01_ but can check the Underworld dvd what was the program I used, wait

14:12 jnettlet yeah just the program. I can look up the process later if I need to.

14:13 mk01_ ok, must have been Mac's original DVDMaker

14:14 mk01_ (no longer shipped by Apple since 10.7)

14:15 mk01_ yes, bunch of .xml tmp files around the iso all having just header : Mac OS X

14:15 mk01_ and This resource fork intentionally left blank

14:15 mk01_ is that info any use for you ?

14:17 jnettlet okay...basically I need to find an OSS way to rip the dvd's. fair enough. Like I said I needed to do that regardless

14:17 mk01_ and look like each component is exported separatelly (into same folder as .iso is saved)

14:18 mk01_ ok, this is structure which plays nicely when ISO opened under XBMC

14:18 mk01_ http://paste.ubuntu.com/12192126/

14:18 mk01_ can move the extra data and retest

14:20 mk01_ menu opens in SW, if I choose a track / section, it opens in VPU

14:21 mk01_ and the files are each about 300k - and that's everything I know about DVDs and how to play them

14:22 mk01_ ((and not DVDMaker, but iDVD was the official application from Apple))

14:24 mk01_ (((but then Steve Jobs decided DVD/BR are topic of 20th century - and removed DVD readers from hw)))

14:26 mk01_ ok, If I move the files menu opens, on click to content xbmc SIGV

14:27 mk01_ but thirt is without MENU plays just by clicking on .iso

14:27 mk01_ with subs / etc

14:27 jnettlet okay thanks. noted

14:28 mk01_ I know, useless, right ? (will keep you in peace, don't worry) :)

21:56 R0nd weird, I can't get a usb sdcard reader working

21:56 R0nd do you think it's a power issue or software?

21:58 R0nd the led on the card reader doesn't light up, but it worked on every system I've tried before, even on my phone

10:09	topi`_	rabeeh: do you offer a Extended temp variant of the HB Edge? At least one component is critical - the eMMC (we have first hand experience of MLC flash corruption when doing writes at sub-zero temperatures and the component being of the commercial variant)
10:10	topi`_	if the Edge does not offer such a variant, then maybe we would need a custom batch of boards with such an eMMC in place?
11:05	rabeeh	topi`_: yes there are extended variants too
11:06	rabeeh	the dual and quad devices are only extended and beyond; we don't use commercial versions of them
11:06	topi`_	right.
11:06	rabeeh	the eMMC device by default is a Samsung 4GByte device.
11:06	topi`_	I'm not worried about the i.MX6, I'm worried about the eMMC
11:06	topi`_	:)
11:06	topi`_	even if the processor fails to start when it's too cold, that's not destructive, however, if the eMMC gets destroyed then that means a field trip for our ops
11:07	topi`_	a few months back we checked the different MMC vendors, and the extended temp variants are usually significantly more expensive than the commercial ones
11:09	rabeeh	KLM4G1FEPD
11:10	rabeeh	^^^ This is the part we are using.
11:18	topi`_	rabeeh: got it. Weird thing is, the codes about that part have nothing that describes the temperature grades for that chip...
11:21	topi`_	actually it might be that all Samsung eMMC parts are for T -25 .. 85
11:21	topi`_	I found one Samsung eMMC with very similar code that is -25..85
11:21	rabeeh	i have the spec somewhere.
11:24	topi`_	that chip seems to be quite new, sampling Q1'15
11:24	topi`_	maybe it works with eMMC 5.0, however I guess the i.MX6 SD controller won't talk MMC5.0?
11:35	topi`_	on Samsung's eMMC site, I see only -25..85 C devices and the -40..105 devices that are labeled "automotive"
11:35	topi`_	so maybe that eMMC is just fine for our purposes
11:37	jnettlet	topi`_, find yourself an industrial freezer and get to testing :)
11:41	mk01_	.... I can provide testing to -74C (if would be really needed)
11:41	jnettlet	mk01_, are you going on an Artic mission?
11:42	topi`_	jnettlet: here in Finland, we just wait for the winter to come, and then can easily test at below -25C :)
11:43	jnettlet	sure, bottle of vodka, long cable, and a sauna. You can test both extremes and enjoy yourself doing it :)
11:44	mk01_	jnettlet: no, but neighbour's daughter married a penguin
11:44	mk01_	;)
11:44	jnettlet	Tux?
11:44	jnettlet	:_
11:44	mk01_	pengu
11:44	mk01_	yup
11:46	mk01_	((-74C is paper wise needed to store bio samples over 4h. so it's not that uncommon to find (those freezers))
11:47	topi`_	just pour a bottle of liquid CO2 and you have -78C :)
11:47	mk01_	yes, allowed. up to 4h
11:47	mk01_	as said :))
11:48	jnettlet	dry ice works as well
11:48	topi`_	installing python deps via pip on i.MX6Q ... it takes quite a lot of time :/
11:49	topi`_	unfortunately none of these seem to be using multiple cores
11:49	topi`_	not even the compile phase; it just launches cc sequentially
11:52	mk01_	jnettlet: have you changed in any way the g2d part in videocodecimx ?
11:52	topi`_	rabeeh: are there any i2eX microSOM's that would offer 2GB of DRAM? it seems the imx6 Dual is only available with 1GB ram?
11:52	jnettlet	mk01_, yes
11:53	mk01_	a easy ones or more rewrite ?
11:53	mk01_	a = an
11:53	jnettlet	a bit of both.
11:53	jnettlet	I will post them when I am doine
12:00	mk01_	because I finished rewriting the general part of CIMXContext (it doesn't need stop/start anymore and fb reconf runs on the fly). I would send that as 2nd PR to upstream, but if you rewrote that other part, I will wait and sync with your result...
12:01	mk01_	makes sense, or ?
12:01	jnettlet	no go ahead and I will rebase on your patches. I am pretty overloaded right now.
12:01	mk01_	ok, no pushing
12:01	jnettlet	Although I think I have fixed the bottleneck using the VPU -> IPU direct rendering
12:05	mk01_	ok, so you removed the circle around *kodi (to get the data from line 1 to line 2?) ?
12:06	mk01_	do you have any draft somewhere, to let me see what will still be valid from the code ?
12:35	mk01_	jnettlet: did you give any thoughts to the PAL/NTSC setup question ? I was going through the logs on from #openelec and you was asking if I PR-ed the fractional modes - basically the commit on the side of xbmc would be 4 lines - so nothing to send in "advance"
12:42	jnettlet	mk01_, lets go with what you have for now. If we decide we need to tweak it we can work from that.
12:42	mk01_	ok
12:42	jnettlet	like I said before, this will be revisited when we switch to KMS so I don't want to dump a bunch of time into it now.
12:43	mk01_	then in the g2d section I just fixed the align on "zooming" out of screen
12:43	jnettlet	are you scaling up and and then cropping?
12:43	mk01_	I cpuldn't find any other misalignments (tested 1600x900 too, 1440x900 and 1366x768
12:44	jnettlet	I actually found one yesterday. Very strange. It was at 1080p but the video was low quality, like DVD resolution.
12:44	mk01_	jnettlet: yes, that one
12:44	mk01_	^^^^ to answer your scale + scrop
12:45	jnettlet	there the entire picture had that wrapping even without the overlay being shown
12:45	jnettlet	okay, if you post your pull request I will pull it down and test later.
12:46	mk01_	jnettlet: what means: "had that wrapping ... " ?
12:47	mk01_	ah maybe I know what you mean
12:47	jnettlet	where the bottom of the picture was wrapping around the framebuffer and being shown at the top. It was flickering so I assumed the pageflipping was wrong
12:47	mk01_	yep
12:47	mk01_	ok
12:47	mk01_	this is what I was telling you before. just coould explain.
12:47	jnettlet	was strange it showed up on normal video playback
12:47	jnettlet	okay
12:48	jnettlet	is this all pushed to your kodi on github?
12:48	mk01_	nope, just cleaning the code.
12:49	mk01_	20m
12:49	mk01_	ok what I want to say -
12:49	mk01_	I can see that too. but only IF doublerate deint is on
12:50	mk01_	and the content has actually no top/bottom info
12:50	jnettlet	I am pretty sure this was progressive, but it is possible it was interlaced.
12:50	mk01_	so all is NONE
12:50	jnettlet	just some random internet video
12:51	mk01_	ok, to be even more clear - content with flag Interlaced, but no frames other from VPU_FIELD_NONE
12:51	mk01_	+ active deint and choosed doublerate
12:52	mk01_	this is when It happens always (for me)
12:53	jnettlet	fun
12:54	mk01_	can you also test some .iso images of DVDs with VPU decoding ?
12:55	mk01_	by default it is forced to ffmpeg, but I have great success with VPU decode. just mine 'test' pool of such media is TWO
12:55	mk01_	so great success = 100% of two files
12:56	jnettlet	mk01_, okay that should work if decss can keep up.
12:56	jnettlet	or are these already stripped?
12:57	mk01_	mmnt
13:05	mk01_	ok, not sure I 'm able to give a proper answer
13:05	mk01_	but the content seems to be properly unpacked , yes
13:05	mk01_	((that's why I'm asking if you can check))
13:08	jnettlet	sure, my wife wants me to rip a bunch of her old DVD's anyways
13:59	mk01_	jnettlet: here it is - https://github.com/xbianonpi/xbian-sources-xbmc/commit/ff9f265c146cddb34bbf22cdff08c5c8d9ba33e7 - but it is absolutely unreadable even as WIP. so have to split it into at least two hunks. one with actual code changes, the other the rest (mostly delete + reorder). later today. ok?
14:00	jnettlet	mk01_, sure no rush. Like I said I am pretty slammed. Just trying to make sure we keep going in a forward direction :)
14:09	mk01_	jnettlet: here is the ffmpeg force so you don't have to look for it https://github.com/xbianonpi/xbian-sources-xbmc/commit/707beaa356efafa03f1de9aaa1c6d4f9e29bc8da
14:09	mk01_	(DVDs)
14:09	jnettlet	mk01_, sure which program are you using to image the DVD's?
14:11	mk01_	that's the thing I explained - I have just two DVDs. otherwise all from iTunes.
14:12	mk01_	but can check the Underworld dvd what was the program I used, wait
14:12	jnettlet	yeah just the program. I can look up the process later if I need to.
14:13	mk01_	ok, must have been Mac's original DVDMaker
14:14	mk01_	(no longer shipped by Apple since 10.7)
14:15	mk01_	yes, bunch of .xml tmp files around the iso all having just header : Mac OS X
14:15	mk01_	and This resource fork intentionally left blank
14:15	mk01_	is that info any use for you ?
14:17	jnettlet	okay...basically I need to find an OSS way to rip the dvd's. fair enough. Like I said I needed to do that regardless
14:17	mk01_	and look like each component is exported separatelly (into same folder as .iso is saved)
14:18	mk01_	ok, this is structure which plays nicely when ISO opened under XBMC
14:18	mk01_	http://paste.ubuntu.com/12192126/
14:18	mk01_	can move the extra data and retest
14:20	mk01_	menu opens in SW, if I choose a track / section, it opens in VPU
14:21	mk01_	and the files are each about 300k - and that's everything I know about DVDs and how to play them
14:22	mk01_	((and not DVDMaker, but iDVD was the official application from Apple))
14:24	mk01_	(((but then Steve Jobs decided DVD/BR are topic of 20th century - and removed DVD readers from hw)))
14:26	mk01_	ok, If I move the files menu opens, on click to content xbmc SIGV
14:27	mk01_	but thirt is without MENU plays just by clicking on .iso
14:27	mk01_	with subs / etc
14:27	jnettlet	okay thanks. noted
14:28	mk01_	I know, useless, right ? (will keep you in peace, don't worry) :)
21:56	R0nd	weird, I can't get a usb sdcard reader working
21:56	R0nd	do you think it's a power issue or software?
21:58	R0nd	the led on the card reader doesn't light up, but it worked on every system I've tried before, even on my phone