08:12 | ralix | morning |
08:19 | wumpus | in case anyone's interested in reverse engineering the gpu in the cubox and helping write an open source driver, there's a project for that now: https://github.com/laanwj/etna_viv/wiki |
10:01 | rabeeh | wumps: hi |
10:02 | rabeeh | wumps: your github caught my eyes :) |
10:22 | wumpus | rabeeh: hey |
10:23 | wumpus | mind that I don't have a cubox myself yet (but another device with a GC800), however the chip should be mostly the same |
10:54 | Thirsty | rabeeh: do you get better support for Marvell these days? |
11:07 | rabeeh | wumps: mmp2? |
11:08 | rabeeh | Thirsty: I get support |
11:08 | rabeeh | Thirsty: r u referring to the court thing? |
11:08 | wumpus | just the Vivante stuff (2D/3D rendering engine) |
11:09 | rabeeh | the gc800; is it a Marvell based device? i recally mmp2 has gc800 inside |
11:10 | rabeeh | wumpus: the vivante gpu is in lots of SoCs today; I think the freescale imx are using them too |
11:11 | wumpus | mine has a rockchip 2918 soc |
11:11 | wumpus | afaik it's marvell armada, rockchip 2918 and indeed the freescale i.mx6 that has vivante.. but there's probably more |
11:12 | rabeeh | there are few more. |
11:12 | rabeeh | what's the idea behind reverse engineering? |
11:12 | rabeeh | the kernel driver is open source |
11:13 | wumpus | open source userspace |
11:13 | rabeeh | (i know) |
11:13 | rabeeh | :) |
11:13 | rabeeh | i meant what's the plan? |
11:13 | rabeeh | in what stage are you on this? |
11:13 | wumpus | well what's mainly needed is a shader compiler now |
11:14 | wumpus | the part about how to render is mostly figured out |
11:15 | wumpus | the most difficult thing is that there are zillion different possibilities with the feature bits, so what works on one device may not work exactly the same on another |
11:15 | Thirsty | rabeeh: half a year ago, graphic driver was a problem and getting respons for better ones for Marvell, take a while. I don't know if it's getting better? |
11:15 | rabeeh | Thirsty: Marvell gets the GPU support from Vivante too; that's the main bottleneck |
11:16 | Thirsty | I finally moved to a my new home. So now I got time again to play with the cubox again |
11:16 | rabeeh | :) |
11:16 | rabeeh | congrats |
11:16 | Thirsty | thanks. |
11:16 | rabeeh | rmull: CuBox pro is changing DDR chip from 2Gb to 4Gb |
11:17 | rabeeh | i don't know if you have rework station; but if you do then you can switch the chips |
11:17 | Thirsty | rabeeh: what packages are the chips? BGA or TSSOP? |
11:17 | rabeeh | BGA |
11:18 | rabeeh | it's the standard jedec 96balls 0.8mm pitch devices |
11:18 | Thirsty | my old company had one :( |
11:18 | rabeeh | wumpus: are you familiar with Vivante profiling capability? |
11:19 | Thirsty | I saw yesterday that shesselba did a lot of work to port the kernel to v3.6. |
11:21 | Thirsty | and also getting it in to the mainline kernel |
11:27 | wumpus | rabeeh: yes, somewhat, the chip has various performance counters per module |
11:30 | wumpus | both low level (number of 8-byte quantities read) and higher level (number of primitives rendered) |
11:33 | wumpus | they're read from the kernel (because userspace can only write state) |
11:54 | Thirsty | rabeeh: are there any graphic kernel drivers v3.6 kernel? |
11:59 | Coburn | I await the day I see Portal 3 on the CuBox |
12:00 | Coburn | :) |
12:00 | bencoh | :) |
12:00 | wumpus | hehe |
12:01 | Coburn | Also, was there any intention to port android on the CuBox? |
12:01 | Coburn | I know 2.2 was ported, but that's OLD |
12:02 | Coburn | 4.2.x is where the flavor is at (Jelly Bean) |
12:02 | Coburn | Reason I ask is that there's a console called OUYA that has a Tegra quadcore in it running Android |
12:02 | Coburn | and it's designed like a PS3/Xbox/etc |
12:03 | Coburn | The console is in the works, but it's got a lot of big-names backing it |
12:03 | Coburn | Just thought I'd bring that onto the battlefield, because it would be a CuBox rival when it comes to gaming |
12:05 | rmull | rabeeh: I do have rework capability - the cubox pro is the only one that will be upgradeable? I have the original non-pro cubox |
12:08 | Coburn | I thought I heard rumors that the original CuBox had 2GB DDR3 RAM, but only 1GB was accessable? |
12:08 | rabeeh | Coburn: the board supported 4Gb devices (4Gb x 4 devices = 2GB) |
12:08 | Thirsty | Coburn: that would be great |
12:11 | Coburn | I don't understand what you mean by 4Gb x 4 |
12:11 | Coburn | I thought that would be 16GB? |
12:11 | Coburn | Do you mean 4Mbits? |
12:11 | Coburn | 4Gbits* |
12:11 | wumpus | Gb not GB :p |
12:11 | Thirsty | 16Gbit / 8 bit = 4GByte |
12:11 | Thirsty | 16Gbit / 8 bit = 2GByte |
12:12 | Coburn | Thirsty: that does not compute |
12:12 | Coburn | you mean 32Gbit / 8 = 4Gbyte |
12:12 | Coburn | ? |
12:13 | Thirsty | yep, I was to fast :) |
12:13 | Coburn | So in the original cubox case |
12:13 | Coburn | It's a 8Gbit RAM chip? |
12:13 | rabeeh | yes |
12:13 | rabeeh | four devices each 2Gb |
12:13 | rabeeh | 4x2Gb = 1GB :) |
12:13 | Coburn | Ohhhhhhhh |
12:14 | rabeeh | now on the pro- |
12:14 | rabeeh | 4x4Gb = 2GB |
12:14 | Coburn | Now it clears up. |
12:14 | Thirsty | rabeeh: now you want to make a 4gbyte version? |
12:14 | Coburn | Damn Gbit terms, can't they just use plain english? |
12:15 | Coburn | I'd upgrade the CPU before hitting 4Gytes |
12:15 | Coburn | of RAM |
12:15 | Coburn | I mean, wouldn't it be a little wasteful having a single core running with 4GB of RAM? |
12:15 | Coburn | Also, rabeeh, what does dual issue mean? Two instructions being fired at once? |
12:16 | rabeeh | 4GB is a waste on any 32bit architecture |
12:16 | Thirsty | I think a lot of server application a lot of memory is usefull. |
12:16 | rabeeh | dual issue mean two instructions at once (if they don't conflict from resource point of view) |
12:17 | rabeeh | but this would be ARMv8 (which is 64bit) |
12:17 | rabeeh | in general more memory can be used on ARM once using LPAE |
12:17 | Coburn | 64bit ARM? Count me in |
12:17 | rabeeh | Coburn: still no devices available on this; it's a pretty new architecture |
12:18 | rabeeh | LPAE in ARM is 40bit physical addressability |
12:18 | wumpus | also I wonder if the focus on 64 bit and high performance will start to cancel out the power advantage ARM has compared to Intel |
12:18 | rabeeh | wumpus: it doesn't |
12:18 | rabeeh | it's a RISC vs. CISC battle |
12:18 | wumpus | cool |
12:18 | Coburn | Intel Atoms compared with ARM are dead in the water |
12:19 | Coburn | I mean, the Atom gets shat on by ARM |
12:19 | rabee | 12:19 * rabeeh agrees |
12:19 | Coburn | I had a board that had a dualcore with HyperThreading Atom 330 |
12:19 | Coburn | When it did USB I/O or connect a SSD to it |
12:19 | Coburn | it bogged down |
12:20 | Coburn | any major I/O = lag |
12:20 | rabeeh | USB 2.0? |
12:20 | rabeeh | or USB 3.0? |
12:20 | Coburn | yeah. |
12:20 | Thirsty | I am not so sure about it. I reed a artical about it. Atom Z2760 vs Tegra 3. Dutch: http://nl.hardware.info/reviews/3740/intel-atom-z2760-vs-nvidia-tegra-3-review-hoe-presteert-windows-8-op-x86-en-arm |
12:21 | Coburn | Intel is a sitting duck. |
12:21 | rabeeh | Coburn: lots of people thought the same :) |
12:21 | rabeeh | i think there will be a fight; no one will know the results yet. |
12:21 | Coburn | I have a ODROID-U2, which is a quadcore Exynos 1.7Ghz PC (same spec as the Galaxy note 2) and it's a ARMv7 |
12:21 | Coburn | It flies |
12:21 | Coburn | Runs Ubuntu 12.10 quickly |
12:22 | Thirsty | I also was looking a that one. |
12:22 | Coburn | Be warned |
12:22 | Thirsty | Too bad, i didn't had 1gbit lan |
12:22 | Coburn | the base unit doesn't come with AC adapter |
12:22 | Coburn | that's $20 extra |
12:22 | rabeeh | it's pretty nakes |
12:22 | rabeeh | it's pretty naked |
12:22 | Coburn | and debug board is $15 extra |
12:22 | rabeeh | i saw you were using the heatsink with it? |
12:23 | Coburn | you saw my youtube video? |
12:23 | rabeeh | yes |
12:23 | Coburn | Yeah, it comes with heat sink |
12:23 | rabeeh | Aluminum heatsink? |
12:23 | Coburn | something like that |
12:23 | rabeeh | is it hot? bearable? |
12:23 | rabeeh | cold? |
12:23 | Coburn | that acts like a stand, and yes, I compiled a kernel on it and it was just warmer than my skin |
12:23 | rabeeh | i mean after working for 1hr doing heavy stuff |
12:23 | rabeeh | that's ok |
12:23 | Coburn | Haven't tested that, but does compiling a kernel count? |
12:24 | rabeeh | make -j4? |
12:24 | Coburn | Seemed to leave one core idle though |
12:24 | Coburn | just sudo make |
12:24 | Coburn | forgot that param |
12:24 | rabeeh | files from SD? SATA? |
12:24 | Coburn | eMMC and uSD onboard |
12:24 | rabeeh | i typically use 'sar' |
12:24 | rabeeh | oh; those are really slow |
12:25 | Coburn | No SATA I don't think... |
12:25 | Coburn | Uh |
12:25 | rabeeh | probably your processors are waiting for I/O to finish |
12:25 | Coburn | well actually |
12:25 | Coburn | sorr |
12:25 | Coburn | y |
12:25 | Coburn | USB SSD |
12:25 | Coburn | in a SATA2 caddy |
12:25 | rabeeh | run 'make -j4' and in parallel run 'sar -du 5 1' |
12:25 | Coburn | what's sar do? |
12:25 | rabeeh | gives stats on lots of things |
12:25 | Coburn | okay, noted |
12:26 | rabeeh | http://linux.die.net/man/1/sar |
12:26 | rabeeh | sar -du 5 1 will collect cpu utilization for 5 seconds and report |
12:26 | Coburn | They say it uses uBoot to start up |
12:26 | Coburn | but I cannot see where the uImage is |
12:26 | Coburn | just zImage |
12:26 | Coburn | so I think it's Samsung bootloader only |
12:26 | Coburn | Supervivi or something like that |
12:26 | rabeeh | i hate those custom boot loaders |
12:27 | Coburn | Well, it's handy |
12:27 | Coburn | just replace the zImage and off you go |
12:27 | Coburn | however, I could be wrong; I don't have debug board so I can't tell you if it is uBoot |
12:27 | Coburn | Gotta order one |
12:28 | Coburn | can uBoot boot zImages? |
12:28 | Coburn | Or only uImage? |
12:30 | Cobur | 12:30 * Coburn pokes rabeeh |
12:30 | Coburn | Looks like I killed the channel. |
12:31 | bencoh | yeah, you silly killer. |
12:32 | Coburn | However, one thing the ODROID-U2 lacks is the support base |
12:32 | Coburn | the support forums are horrible |
12:33 | Coburn | phpBB2 and the people there are questionabl |
12:33 | Coburn | e* |
12:33 | Coburn | like they don't read the FAQs, don't use search, etc. |
12:34 | Coburn | Not bashing them, but it's patchy |
12:40 | rabeeh | here |
12:40 | rabeeh | u-boot can boot zImage |
12:41 | rabeeh | Coburn: where can i get more information on the lcd register issue thing? |
12:42 | rabeeh | have a pointer? or something? |
12:42 | Coburn | [21:40:47] < rabeeh> u-boot can boot zImage <- Aha! So that's what happens then! |
12:42 | Coburn | rabeeh: your best info would be _rmk_, but he was sent to exile, no? |
12:43 | rabeeh | yea |
12:43 | rabeeh | i don't know what happened there |
12:44 | Coburn | I think he just got pissed off to the extent that he didn't want to talk to you anymore |
12:44 | Coburn | Not blaming anyone |
12:44 | Coburn | Just observation |
12:44 | Coburn | He could have been in a ragey mood too. |
12:44 | Coburn | bribe him back and interrogate him :P |
12:45 | Coburn | Ok |
12:45 | Coburn | [21:45:02] NickServ [NickServ@services.]: Last seen : Jan 03 12:00:37 2013 (4 days, 23:44:31 ago) |
12:45 | Coburn | He's been MIA |
12:56 | rabeeh | i sent him an email clearing things up |
12:57 | Coburn | That's good |
12:57 | rabeeh | well.. that was 10minutes after he left; never got anything back |
12:57 | Coburn | :( |
12:57 | Coburn | If I had the tools |
12:58 | Coburn | I'd be able to give you dumps of the registers when it "derps" |
12:58 | Coburn | derp = goes stupid |
12:58 | Coburn | It's mostly present in GeeXbox. |
12:58 | rabeeh | i still don't understand what the issue is |
12:58 | rabeeh | oh; so it's on geexbox |
12:58 | rabeeh | how do you reproduce? what do you see? |
12:59 | Coburn | dbsx hasn't experienced it, but I have (he's got one of the first cuboxes). |
12:59 | Coburn | From what I understand |
12:59 | rabeeh | dbsx is our Aussie distributor :) |
12:59 | Coburn | I boot up GeeXbox |
12:59 | rabeeh | he has lots of CuBoxes :) |
12:59 | Coburn | Yup |
13:00 | Coburn | My cubox I had is at my brothers, but what he does is |
13:00 | Coburn | Boot up CuBox -> select video -> it buffers (WORKING....) and then the screen flickers as the video resolution is set |
13:00 | Coburn | sometimes, the screen will flicker, and the video will play back, but then the overlay with time, clock and controls will be skewered |
13:01 | Coburn | This is Standard Definition DVD Rips. a mixture or standard and high profile. |
13:01 | rabeeh | skewered ? |
13:01 | Coburn | when the system is glitched and the register is malfunctioning, a diagnoal line exists |
13:01 | rabeeh | completely corrupted? |
13:02 | Coburn | let me get you a photo. |
13:02 | rabeeh | oh great; photo :) |
13:03 | Coburn | come on twitter, you're so slow |
13:04 | Coburn | rabeeh: https://twitter.com/coburn64/status/278649358164058112/photo/1 |
13:04 | Coburn | That explains it quite well |
13:04 | rabeeh | oh |
13:04 | rabeeh | so the RGB pitch got corrupted |
13:04 | rabeeh | so good; it's an LCD controller issue and not a GPU |
13:04 | Coburn | ...pitch? |
13:05 | rabeeh | if you look the line meats on the middle at the bottom |
13:05 | rabeeh | i mean the diagonal line |
13:05 | rabeeh | that means the pitch is missed in a single pixel |
13:05 | Coburn | Ok |
13:06 | Coburn | And with this, you can file a defect with Marvell? |
13:06 | rabeeh | :) |
13:06 | rabeeh | it's a driver issue that I can probably fix |
13:06 | rabeeh | but i need to recreate |
13:06 | rabeeh | well.. one thing |
13:06 | rabeeh | how th resolution changes? did u set the resolution to change upon video resolution? |
13:07 | Coburn | I don't know. The screen just flickers |
13:07 | rabeeh | i mean if it's 720p movie; did u set it to change the gui resolution? |
13:07 | Coburn | Could it be just XBMC warming up? |
13:07 | Coburn | I don't think so |
13:07 | Coburn | I can't remmeber |
13:07 | rabeeh | what happens once you exit from the movie? |
13:07 | Coburn | stays skewered |
13:07 | rabeeh | does it go well back again? |
13:07 | rabeeh | ok. |
13:07 | Coburn | need to power cycle it |
13:08 | Coburn | xorg reboot may fix it... |
13:08 | Coburn | not sure. again, I need to get a cubox off dbsx since mine was sold to my brothers |
13:08 | rabeeh | it's probably something went bad when enabling the video plane; something is messing with the graphics overlay |
13:08 | Coburn | Can you look into it? |
13:09 | rabeeh | yes. but first i need to reliably reproduce it |
13:09 | Coburn | I'll let you know this |
13:09 | Coburn | Got Windows? Try ripping a DVD and then encoding using handbrake using high profile mp4 |
13:09 | Coburn | then put that on usb hdd/stick and play back |
13:09 | Coburn | stop the video, play it agian |
13:10 | Coburn | do that like 20 times and see if you can get it to play around |
13:10 | rabeeh | is this why _rmk_ was angry about? |
13:11 | Coburn | somewhat |
13:11 | Coburn | if i recall, you were saying "no, it can't happen" and he said "I can replicate it" |
13:11 | Coburn | i could be wrong. don't quote me on that |
13:12 | Coburn | but anyway. If we can get this issue fixed, it's all for the best. I'd be willing to provide register dumps and such |
13:12 | rabeeh | Coburn: if you were able to reproduce (or anyone on the channel); there is a 'cat /proc/mv_lcd0' that dumps all registers on the LCD controller |
13:12 | rabeeh | it would be helpful getting a dump before and after the issue happens |
13:13 | Coburn | beautiful |
13:13 | Coburn | I will do that ASAP |
13:13 | Coburn | don't forget to put my name in the "bugfix credits" |
13:13 | Coburn | ;) |
13:15 | rabeeh | :) |
13:15 | Coburn | luckily, I have a EXACT uSD image that my brother has of the XBMC setup |
13:16 | Coburn | so I can deploy it and saves me the hassle of setting things back up |
13:24 | dbsx | Coburn: order the CuBox and it can be there on Thursday (or do you need another credit?) |
13:28 | rabeeh | Coburn: The Smurfs? |
13:32 | Coburn | dbsx: I'm still waiting on payment from my brother. |
13:32 | Coburn | rabeeh: yeah, I was tasked to rip DVDs that my brother's nieces watch, so that video was the first one that glitched |
13:32 | Coburn | the next glitch happened on Shrek |
13:33 | jnettlet | doing graphics development I have learned to treat the saved screenshots the same as a doctor or lawyer would. complete confidentiality and zero personal judgement |
13:34 | bencoh | :) |
13:35 | Coburn | anyway, gotta get to bed. parents bugging me to turn lights out and stuff. |
13:36 | jnettlet | have sweet dreams of small blue blondes ;-D |
13:36 | Coburn | oh please |
13:36 | Cobur | 13:36 * Coburn cringes |
13:36 | Coburn | dbsx: I'll call you tomorrow re cubox. |
13:37 | Coburn | for now, I gotta get some shuteye, g'nite. |
13:47 | diget | yeah! Finally the waiting has an end. My cubox just arrived this moment \o/. Looks really nice. |
14:03 | Thirsty | I can try some files. |
14:07 | Thirsty | I installed the geexbox-devel-20121206-r15609.cubox.tar.bz2 version on my SDcard |
14:19 | Thirsty | I see that solid-run has a deb package repository. I also added to source.list. Which distrobution do I need? I currently running Debian Wheezy armhf. |
14:20 | Thirsty | I get the following errors: libbmm-dev : Depends: libc6 (>= 2.15-0ubuntu8) but 2.13-37 is to be installed |
14:21 | Thirsty | I am up-to-date with debian |
14:27 | Thirsty | Ok, I found it. http://www.solid-run.com/phpbb/viewtopic.php?f=11&t=691&p=5988&hilit=source.list#p5988 |
22:02 | niteye | how to tell if you have cubox v1 or v2 and whats the difference ? |
22:04 | neofob | niteye: the v1 cubox has the serial usb port above the esata port |
22:04 | niteye | ah okay |
22:05 | niteye | frustrating, i dont have a micro-usb B cable heh |
22:05 | neofob | you should, most cellphone these days use it to charge |
22:06 | niteye | my phone is from before the age of connecting phones via USB, ill have to scrounge in my sister's room tomorrow |