=== lilstevie|ZNC is now known as lilstevie | ||
=== asac_ is now known as asac | ||
=== panda is now known as Guest38046 | ||
MrCurious | persia: are you around? | 04:05 |
---|---|---|
MrCurious | i think it was you who reccomended that i use the ubuntu opencv linraries (from apt-get) rather than building them myself. do you know if for pandaboard, the ubuntu opencv libraries have any DSP or 3d optimizations in them | 04:06 |
Martyn | they do not, that I am aware of | 04:06 |
MrCurious | curses! | 04:06 |
MrCurious | wonder if anyone is working on hardware acceleration for opencv on pandaboard | 04:08 |
Martyn | Well, there isn't all that much in there to help with accel | 04:08 |
Martyn | It's not exactly like you can use CUDA on an omap 4400 | 04:08 |
Martyn | openCL maybe? | 04:09 |
Martyn | that uses the SGX accel | 04:09 |
MrCurious | looks like opencl has not yet come to the pandaboard | 04:12 |
MrCurious | hmmm OpenCL implementation is in work at this time and from what I understand it | 04:13 |
MrCurious | will likely be made available in 2H11. Andrew is at Imagination Technologies | 04:13 |
MrCurious | ( the graphics IP vendor for OMAP4 SoC) i.e, he has access to the inside | 04:13 |
MrCurious | implementation he he is able to do such posts. | 04:13 |
MrCurious | wonder what 2H11 means? | 04:13 |
MrCurious | 2nd quarter 2011 | 04:13 |
MrCurious | http://withimagination.imgtec.com/news/powervr-sgx-cores-get-opencl-conformance/ | 04:19 |
MrCurious | looking like you are right about opencl | 04:21 |
MrCurious | its just a question of can they be had | 04:21 |
persia | MrCurious, I don't know of anyone working on optimisation of OpenCV (except perhaps you). | 05:16 |
MrCurious | i am not working on, just inventorying what is available to me | 05:20 |
MrCurious | i am curious if anyone has a pandaboard and a web cam(running ubuntu or android) if they could check teh FPS they get with mplayer on /dev/video0 | 05:21 |
MrCurious | would love to know if i have a less than optimum setup 3 web cams varying 8 FPS to 12 FPS | 05:22 |
persia | Looks like OpenCL isn't really well-supported right now. There's a khronos-opencl-headers package, but it doesn't seem to be something that has dependencies. | 05:23 |
MrCurious | no library package for those headers? | 05:23 |
persia | Only the nVidia CUDA implementation | 05:23 |
MrCurious | and my guess is that is not meant for pandaboard | 05:24 |
persia | And from what I can tell, that's not being distributed as part of Ubuntu currently. | 05:24 |
persia | Heh, no. | 05:24 |
persia | The headers are from http://www.khronos.org/registry/cl/ , but from what I can tell, it needs N implementation libraries. | 05:25 |
persia | Meaning that there's probably lots of integration and interoperability work to be done as the implementations become available and packages are created. | 05:26 |
persia | For the pandaboard, you'll probably get the fastest throughput with video capture using gstreamer. | 05:28 |
persia | I believe TI has some optimised gstreamer input plugins in their PPA. | 05:28 |
persia | (although I may be mistaken: I haven't tried playing with video in) | 05:28 |
MrCurious | gstreamer... isnt that for working the DSP's? | 05:29 |
MrCurious | not something to speed up USB ports | 05:29 |
persia | gstreamer is an A/V programming framework: I believe that it helps deal with onboard processing once it's come over USB. | 05:32 |
persia | If you think 8-12 FPS is limited by USB, then it probably won't help. | 05:32 |
MrCurious | at this point i am unsure of where its bottlenecked. thats why i seek comparison with other users | 05:35 |
MrCurious | given that my disk was much much slower than gruemaster | 05:35 |
MrCurious | and mine was ssd vs his spinning disk | 05:36 |
MrCurious | panda has gstreamer support? | 05:40 |
persia | Everything has gstreamer support, but TI has some special gstreamer modules to make panda even faster at processing video. | 05:42 |
MrCurious | interesting | 05:43 |
MrCurious | lots of packages come back in a gstreamer search | 05:44 |
persia | Right, but the ones with the panda-specific optimisations are probably those in ppa:tiomap-dev/release | 05:45 |
persia | I *think* you want gst-ducati (but I'm not really sure) | 06:00 |
Martyn | Damnit, I can't get a hold of George in New Zealand | 06:03 |
persia | It's the end of the day there... | 06:03 |
Martyn | yes.. but the earthquake seems to have knocked out phones/cells again | 06:04 |
Martyn | there was a swarm, starting at 1pm (5.5) ending with a pretty solid 6.0 quake at 2:20pm .. it's been a little over two hours | 06:04 |
persia | I seem to have some (sporadic) traffic from .nz hitting my router: I presume the lines are up and congested. | 06:09 |
Martyn | .nz is big enough to be up away from Christchurch | 06:12 |
Martyn | can you tell if there is much (if any) traffic coming from there? | 06:12 |
persia | Nope. I have insufficient knowledge of the relation between network topology and geography in that region. | 06:12 |
persia | I'm probably currently biased, but I'd presume that a series ending in 6.0 isn't likely to have been catastrophic, although that means nothing in terms of individuals. | 06:14 |
Martyn | Yep, and what's bad is that NZ has many, many buildings that just aren't up to earthquake codes | 06:14 |
Martyn | they happen, and lately frequently, but in the past not very often | 06:15 |
persia | That could make it much worse then. | 06:15 |
Martyn | Can .. of course since the big quake four months ago, there has been a call to up the quality of construction (and shore/replace current structures) | 06:17 |
Martyn | but it's only been four months | 06:17 |
persia | Code compliance requires a generation of buildings, which can be fairly long in places with heritage regulations. | 06:18 |
=== tonu__ is now known as tonu | ||
garagoth | Good morning | 10:25 |
garagoth | I have a problem with mu Natty on BB-xM | 10:27 |
garagoth | (again) | 10:27 |
garagoth | After reboot it does not bring up USB and network | 10:27 |
garagoth | also, bridgedriver (for DSP) seems to be missing... | 10:30 |
garagoth | any hints? | 10:33 |
GrueMaster | What rev beagleXM do you have? | 10:33 |
garagoth | C | 10:34 |
garagoth | it was working before reboot (except dsp) | 10:34 |
GrueMaster | Yep. Known issue. We have a new kernel for this. Give me a sec and I'll paste the link up. | 10:35 |
garagoth | and kernel during startup correctly identifies my board (u-boot does not) | 10:35 |
garagoth | GrueMaster: Is this a second patch for Natty on xM rev B, C then? | 10:35 |
GrueMaster | Yes. | 10:36 |
garagoth | But it started up correctly when I rebooted after headless install... ? | 10:36 |
GrueMaster | https://wiki.ubuntu.com/ARM/OmapNetbook | 10:37 |
GrueMaster | The update is at the bottom of the wiki page. | 10:37 |
garagoth | I powered it off for weekend and it did not start. | 10:37 |
GrueMaster | Instructions are the same for headless. | 10:37 |
garagoth | GrueMaster: Is this the same updated image I applied at hm, thursday or friday, when I reported problems with headless install here? | 10:38 |
GrueMaster | Maybe. I have lost track, as I had to fly to London for meetings. | 10:38 |
GrueMaster | I can look through the backscroll to check. | 10:39 |
garagoth | Because headless install was failing (no network), then applied your (and rsalveti) patch, went fine, reboot, still all fine... now not fine, after halt & poweroff | 10:39 |
GrueMaster | Odd. Same kernel or did it revert? | 10:40 |
garagoth | The patch is exactly the same, md5 identical. | 10:41 |
garagoth | Hm, dunno. | 10:41 |
garagoth | Should be the same? | 10:41 |
garagoth | How can I check it? | 10:41 |
GrueMaster | Here is my md5sums: | 10:43 |
GrueMaster | 6b627ef6dbbca9b9f8212a512fcd9561 uImage | 10:43 |
GrueMaster | 1747c4ab2cc6a9d0710207f48666a91d vmlinuz-2.6.38-8-omap | 10:43 |
GrueMaster | I don't have access to a beagleXM here, so I can't check the kernel signature. | 10:44 |
garagoth | my uImage is different | 10:44 |
garagoth | but I changed boot.script and executed flash-kernel | 10:45 |
GrueMaster | flash-kernel would have changed the timestamp, but nothing else as long as it is using the same vmlinuz. | 10:46 |
garagoth | Hmm! | 10:46 |
GrueMaster | Make sure /boot/vmlinuz links to the vmlinuz updated kernel. | 10:47 |
garagoth | it is. | 10:47 |
garagoth | and flsh-kernel generated uImage with md5 2fb0dd0f4061e4b5c0ff482639d69248 | 10:51 |
GrueMaster | I don't know what to tell you off hand. maybe rsalveti will have some ideas. | 10:53 |
garagoth | what is the kernel module responsible for usb on BB-xM ? | 10:53 |
GrueMaster | Not sure, but I know that it is built in. | 10:54 |
garagoth | ahm. | 10:54 |
garagoth | Then maybe you know something about missing bridgedriver module? | 10:55 |
GrueMaster | Not offhand. | 10:56 |
garagoth | Mm. Thanks. | 10:57 |
GrueMaster | Let me see if my beaglexm is still running at home. If so, I can do a quick test of this kernel. | 10:57 |
garagoth | Sure. | 10:58 |
garagoth | Angstrom boots fine, so not a hardware issue... | 11:03 |
garagoth | GrueMaster: Also, u-Boot still reports Beagle unknown rev 0x02, while u-Boot from Angstrom displays Beagle xM. Kernel displays correctle as bb-xm rev C... | 11:10 |
garagoth | Grr. Why I always make a typo in correctly and usually in typo ... | 11:11 |
GrueMaster | Yes, I don't know if u-boot has been updated upstream yet. | 11:12 |
GrueMaster | Also, I can't get the kernel to work properly with networking either on my older XM, so definately a kernel issue. | 11:13 |
GrueMaster | rsalveti: ^^^ | 11:13 |
=== nielsslot_ is now known as nielsslot | ||
garagoth | rcn-ee: How is your work on 2.6.39 kernel for natty? | 12:33 |
jussi | Does anyone know about the driver for the HDMI expansion card on the freescale iMX53 quick start board? (ie. which driver it is) | 12:37 |
rcn-ee | garagoth, haven't been able to repeat it, tried a bunch of mmc cards, haven't been able to get the mmc -110 on the current image.. | 13:36 |
garagoth | rcn-ee: Also, your latest 2.6.39.1-x1 kernel fails to bring up my USB devices & network interface... | 13:37 |
rcn-ee | are you using my u-boot? | 13:37 |
garagoth | I installed it on natty headless because 2.6.38-8 has same problem | 13:37 |
garagoth | rcn-ee: Not sure... | 13:38 |
rcn-ee | android people where running the same issue.. if you don't have a u-boot that knows about the xM C, for some reason the usb is dead.. | 13:38 |
garagoth | Funny thing is... | 13:38 |
garagoth | that I installed headless natty (after installing patched 38-8 kernel), rebooted, all was fine... then I powered it off, waited 2 days, powered on and no USB | 13:39 |
rcn-ee | so either, something was setup that got lost between reboot's, or the kernel changed.. | 13:43 |
garagoth | Kernel was the same... | 13:44 |
garagoth | But few reboots it was working. It stopped after power off | 13:44 |
garagoth | for few reboots* | 13:44 |
rcn-ee | humm strange.. well got to run to work, will be back in just a bit.. | 13:45 |
garagoth | Mm. | 13:46 |
garagoth | Ok. More strange things | 13:49 |
garagoth | I booted Angstrom (it booted fine, all USB working), then reset (without powering off), swapped cards, booted ubuntu... and all is working again... | 13:50 |
garagoth | rcn-ee: Let me know when you will be back, I need to learn about your uBoot | 14:47 |
rcn-ee_at_work | garagoth, nothing special, i just build it from angstrom's.. | 14:48 |
garagoth | Ok... how to install it? It is "special" as it works... | 14:49 |
rcn-ee_at_work | on your fat partition, just copy it as is. "uboot.bin" would upgrade it.. MLO/X-loader it's usualy a toss up 50/50 you'll need to redo the sd card.. | 14:50 |
garagoth | what would it upgrade? | 14:55 |
garagoth | Maybe I will just copy MLO & u-boot.bin from Angstrom? Will it work, or your files have something better/Ubuntu specific? | 15:05 |
=== prpplague is now known as prpplague^2 | ||
GrueMaster | I would hope that we all are using the same u-boot source tree, instead of one for ubuntu, one for Angstrom, etc. Our released package may be dated, but we should be using the same upstream source. | 15:14 |
rcn-ee_at_work | they are 2011.02 with a few patches: http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/u-boot/u-boot_git.bb?h=2011.03-maintenance#n28 | 15:17 |
=== Jef91|AFK is now known as Jef91 | ||
garagoth | rcn-ee: None of your u-boot.bin files correctly identified my board | 15:52 |
garagoth | while Angstrom u-boot.bin identifies it correctly. | 15:53 |
garagoth | I was checking inages from here: http://rcn-ee.net/deb/tools/beagleboard/ | 15:53 |
rcn-ee_at_work | yeah, sorry those are pure mainline was using them to find a mmc card issue on an old Bx... | 15:54 |
rcn-ee_at_work | for the xM: http://rcn-ee.net/deb/tools/UBOOT/u-boot-beagleboard-2011.02+r75+gitrc7977858dcf1f656cbe91ea0dc3cb9139c6a8cc8-r75.bin | 15:54 |
garagoth | ok, let me try.... | 15:54 |
rcn-ee_at_work | (that's one i built from angstrom's tree and uploaded) | 15:55 |
garagoth | Original Angstrom u-boot.bin file is looking for /boot/uImage from mmc device 0:2 | 15:55 |
garagoth | does your look for /uImage on mmc 0:1 ? | 15:56 |
rcn-ee_at_work | no, i redirect it with: http://elinux.org/BeagleBoardUbuntu#boot.scr_-.3E_uEnv.txt (on my demo images... ) | 15:57 |
brendand | is there a way to get flash working on pandaboard? | 15:58 |
garagoth | mmc0: unrecognised SCR structure version 6 | 16:00 |
garagoth | mmc0: error -22 whilst initialising SD card | 16:01 |
garagoth | mmc_erase: erase error -110, status 0x800 | 16:01 |
garagoth | end_request: I/O error, dev mmcblk0, sector 334793 | 16:01 |
garagoth | This is on your 39.1-x1 kernel | 16:01 |
rcn-ee_at_work | I'd really test another mmc card.. i might have the same one that came from beagle here at work. | 16:03 |
garagoth | rcn-ee: This is different card. | 16:32 |
garagoth | Transcend microsd | 16:33 |
garagoth | 4GB | 16:33 |
garagoth | but there is only one such error and system boots up | 16:33 |
garagoth | on stock card there were hundreds of those... | 16:33 |
=== zyga is now known as zyga-afk | ||
garagoth | rcn-ee: On 2.6.38-8 kernel there are no such errors. | 17:02 |
garagoth | and there seems to be some error with usb0 (network) - interface is up, bot not communication goes through. On 2.6.38-8 all is fine again. | 17:03 |
garagoth | but your u-boot works. | 17:05 |
rcn-ee_at_work | garagoth, the only error i'm getting so far is: "mmc0: unrecognised SCR structure version 10, mmc0: error -22 whilst initialising SD card" but i believe that's related to the expansion mmc setup (it doesn't find it) | 17:05 |
garagoth | What can I say. | 17:06 |
rcn-ee_at_work | the one big difference between my 2.6.39 and the older 2.6.38 is we know have support for this mmc based device: http://boardzoo.com/product_info.php?products_id=1 | 17:08 |
garagoth | Beagle wlan adapter? | 17:09 |
rcn-ee_at_work | same wlan as on the panda | 17:09 |
GrueMaster | beagle has wlan? | 17:10 |
rcn-ee_at_work | expansion | 17:10 |
GrueMaster | ah. | 17:11 |
rcn-ee_at_work | garagoth, i had one of the mmc card's that came with a beagle xm, but i can't even write an image to it without that task hanging at the moment.. here's my pastebin with a good working card: http://paste.ubuntu.com/625969/ | 17:19 |
garagoth | Hm, looks like you have different xM revision | 17:20 |
garagoth | Mine is: [ 0.149444] OMAP3 Beagle Rev: xM C | 17:21 |
rcn-ee_at_work | yeah in the lab here i only have my "xM B" labeled, my C is out in the field with another engineer.. | 17:21 |
NCommand1r | rsalveti: did you file a bug on the OMAP3/4 boards failing to reset with the new images? | 17:26 |
rsalveti | NCommand1r: nops, as I didn't test with the new images yet | 17:26 |
NCommand1r | rsalveti: well I haven't built a new image yet (and the old images are sitll pre-install, but I'm working on it) | 17:27 |
GrueMaster | NCommand1r: what reboot failure? | 17:29 |
GrueMaster | On oneiric? | 17:29 |
NCommand1r | GrueMaster: when you boot a panda with a partition formatted with the new mkdosfs patch, the reset button onthe panda stops working | 17:30 |
GrueMaster | Ah. | 17:30 |
GrueMaster | I wonder if it is similar to the toolchain issue with the omap3 kernel and usb? | 17:30 |
GrueMaster | bug 791552 | 17:31 |
ubot2 | Launchpad bug 791552 in linux "No USB support on beagle/beagleXM" [High,New] https://launchpad.net/bugs/791552 | 17:31 |
NCommand1r | rsalveti: is three a packaging for x-loader/u-boot? I need to create udebs | 19:44 |
=== zyga-afk is now known as zyga | ||
rsalveti | NCommand1r: yup, for both | 20:58 |
rsalveti | NCommand1r: why do you need udeb for them? | 20:59 |
NCommand1r | rsalveti: for d-i | 21:09 |
garagoth | Hmm. I have Beagle-xM trainer board from tin can tools. u-boot correctly detects it. | 21:35 |
garagoth | But, when ubuntu boots, I cannot see i2c bus 2 | 21:36 |
garagoth | only 1 and 3 are visible | 21:36 |
garagoth | any hints? | 21:36 |
garagoth | (Angstrom boots and sees i2c bus 2 out of the box and I am able to communicate on it. So not a hardware issue) | 21:36 |
rcn-ee_at_work | garagoth, i don't think ubuntu's kernel has support for any of the expansion board's.. (the patches angstrom have come with haven't been pushed to mainline either..) | 21:43 |
garagoth | but i2c bus 2 is available on expansion port | 21:43 |
garagoth | beagle trainer board just translates voltage levels | 21:43 |
rcn-ee_at_work | only 1 & 3 are visable because port 2 isn't setup in ubuntu or mainline either: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=arch/arm/mach-omap2/board-omap3beagle.c;h=7f21d24bd437732724a45af9302c50a514f56482;hb=HEAD#l418 | 21:45 |
rcn-ee_at_work | for the expansion boards you need a patch similar to this: https://github.com/RobertCNelson/stable-kernel/blob/master/patches/sakoman/2.6.39/0032-OMAP3-beagle-add-support-for-expansionboards.patch | 21:46 |
garagoth | Just great... | 21:47 |
garagoth | Thanks. will think over it tomorrow. It is already 11 PM here. I need to sleep... | 21:48 |
garagoth | Cheers! | 21:48 |
garagoth | and good night... | 21:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!