| 12:38 | rabeeh | silent today |
| 12:38 | rabeeh | is 2013 going to be a quite year? no flame war? nothing? |
| 12:41 | Coburn | Hopefully the year of the CuBox |
| 12:43 | _rmk_ | "a quite year" ? |
| 12:43 | _rmk_ | quiet? |
| 12:44 | Coburn | as long as 2013 fixes that stupid lcd derp |
| 12:44 | _rmk_ | maybe if more companies hit on bad times. |
| 12:44 | Coburn | where the register gets popped over a barrel |
| 12:44 | Coburn | and causes XBMC stuff getting skewered with that line |
| 12:44 | _rmk_ | you keep going on about that, and I'm just going to keep on telling you that it's a hardware problem which software can't detect |
| 12:45 | _rmk_ | so we can't fix it in software unless Marvell produce an errata for it to say whether it can be worked around |
| 12:45 | Coburn | are we going to get a hardware revision to fix it? |
| 12:45 | _rmk_ | no idea, does Marvell even know about it? |
| 12:45 | Coburn | ah |
| 12:45 | Coburn | someone prod the bastards |
| 12:45 | Coburn | the lawsuit possibly might cripple them though |
| 12:46 | _rmk_ | lawsuit? |
| 12:46 | Coburn | yeah |
| 12:46 | Coburn | see the CuBox site |
| 12:46 | Coburn | some patents got broken |
| 12:46 | Punkley_Chillin | http://www.solid-run.com/media-releases/63-official-statement-regarding-marvell-semiconductors |
| 12:46 | Coburn | I will say though, with galcore, it is way ahead of Smasung chips |
| 12:46 | Coburn | Samsung Mali has no 2D GPU accel |
| 12:46 | Coburn | only in android |
| 12:47 | _rmk_ | oh that. |
| 12:48 | _rmk_ | given the idiocracy with Marvell and their Feroceon stealing the ARM Ltd ARM926 ID and trying to pass itself off as an ARM926 so that "software was compatible" but in reality it had soo many differences it wasn't... I'm not surprised. |
| 12:49 | Coburn | alright |
| 12:50 | Cobur | 12:50 * Coburn is going to bed, so good night. |
| 12:50 | _rmk_ | I think ARM Ltd gave them quite a kicking (privately) for that |
| 12:50 | Coburn | also rabeeh |
| 12:50 | Coburn | can we have CuBox Elite with dualcore CPUs? |
| 12:50 | Coburn | :P |
| 12:50 | Coburn | CPU* |
| 12:50 | Coburn | There's a amada SoC that does that I think |
| 12:50 | Punkley_Chillin | fdual pffft i want quad :) |
| 12:51 | Coburn | how about two quad cores? |
| 12:51 | Coburn | 8 core cubox |
| 12:51 | Punkley_Chillin | 4 quads |
| 12:51 | Coburn | 16 eight-cores |
| 12:51 | Punkley_Chillin | be like the server at my office |
| 12:51 | _rmk_ | coburn: coherency would be a big issue |
| 12:51 | Punkley_Chillin | 32 cores, 256gig memory |
| 12:52 | Coburn | coherency? |
| 12:52 | _rmk_ | EYOURALLSILLY :) |
| 12:52 | _rmk_ | yes, coherency between the cores is very important if you want a working SMP system |
| 12:52 | rabeeh | _rmk_ coherency is a non issue |
| 12:53 | rabeeh | CuBorn: " as long as 2013 fixes that stupid lcd derp" ? |
| 12:53 | rabeeh | wtf? |
| 12:53 | _rmk_ | you can't take two independent ARM CPUs, stick them on a board, and build a SMP system out of that. |
| 12:53 | rabeeh | nop |
| 12:53 | _rmk_ | rabeeh: rubbish, you need coherency for a SMP system, otherwise it's an AMP system. |
| 12:53 | rabeeh | but there are companies building 32 cores |
| 12:54 | Coburn | rabeeh: I mean, I hope to see in 2013 a version of the CuBox without the LCD register issue |
| 12:54 | rabeeh | i know that; i think you will be seeing 32 cores in a single chip |
| 12:54 | dotarra | 12:54 * dotarray takes notes |
| 12:54 | rabeeh | Coburn: which LCD register issue? what did i miss? |
| 12:54 | rabeeh | i know the power management issue _rmk_ mentioned |
| 12:54 | _rmk_ | rabeeh: please read what was being said above before jumping in with silly comments. "how about two quad cores" |
| 12:54 | Coburn | [21:44:14] where the register gets popped over a barrel |
| 12:54 | Coburn | [21:44:30] and causes XBMC stuff getting skewered with that line |
| 12:55 | rabeeh | _rmk_: two quad core is also possible |
| 12:55 | _rmk_ | not as two separate chips |
| 12:55 | rabeeh | you can do coherency between clusters; it's called NUMA |
| 12:55 | rabeeh | not as two separate chips? why not? |
| 12:55 | Coburn | look up something called |
| 12:55 | Coburn | ARM EnergyCore |
| 12:55 | rabeeh | SGI did that 50 million years ago |
| 12:56 | Coburn | they have 4 CPU quadcore addin boards |
| 12:56 | _rmk_ | unlike x86, there's very little support for coherency off chip |
| 12:56 | rabeeh | no; ARM EnergyCore are not NUMA or SMP; they are connected with 10g ethernet fabric |
| 12:56 | _rmk_ | which means you're going to have to keep on flushing the caches, which means poor performance |
| 12:57 | rabeeh | _rmk_: what do you mean? |
| 12:57 | Punkley_Chillin | well look at that ubuntu 12.10 armhf with ubuntu-desktop |
| 12:57 | rabeeh | I/O coherent, CPU coherency has been in ARM for few years |
| 12:57 | _rmk | 12:57 * _rmk_ rotfls |
| 12:57 | rabeeh | Discovery XP does that; i know it's tough to have that |
| 12:58 | _rmk_ | really? Only in a _very_ few Intel IXP chips |
| 12:58 | _rmk_ | and only recently has IO coherency been finding its way into other places |
| 12:58 | Coburn | > intel |
| 12:58 | _rmk_ | and only in the last year has ARM come up with architecting a coherent bus port |
| 12:59 | Coburn | Anyway, I'll leave you guys duke it out. |
| 12:59 | Coburn | G'nite. |
| 12:59 | rabeeh | good night Coburn |
| 12:59 | rabeeh | _rmk_ ARM spreads out architecture license; and companies can implement whatever they want |
| 12:59 | Coburn | before i go, one idea |
| 13:00 | _rmk_ | rabeeh: yes, this is why almost everything we have in the kernel now treats every device as incoherent, except for Intel. |
| 13:00 | _rmk_ | of course you're right. |
| 13:00 | _rmk | 13:00 * _rmk_ goes away. |
| 13:00 | Coburn | _rmk_ said that it's possible to reset the lcd register that gets tangled up and causes XBMC stuff to get skewered. If we know how to access that register, then we can just poke that register and make it into a daemon |
| 13:01 | Coburn | so every 15 mins, it pokes that register if it's not correct, and it fixes it |
| 13:01 | Coburn | but if it's a hardware issue, well, it hurts CuBox somewhat. |
| 13:02 | Coburn | Anyway, bed calls my name. Talk to you all tomorrow. |
| 13:02 | rabeeh | i must have missed that |
| 13:03 | rabeeh | i'm looking at the logs above and can't see that register. can you provide more details? |
| 13:03 | Punkley_Chillin | grrr ubuntu-desktop blank screen, getting there :) |
| 17:39 | michaelcubox | Hello all, anybody already has compiled a 3.7 headless kernel from original source yet? |
| 18:22 | neofob | michaelcubox: no, i'd love to try that |
| 18:23 | neofob | i tried to merge vDorst's 3.5.7 with Linus' master but there are so many conflict so... |
| 18:26 | neofob | also, there is a series of crypto patches from Phil Sutter that need to be applied for mv_cesa |
| 18:26 | neofob | http://thread.gmane.org/gmane.linux.kernel.cryptoapi/7162 |
| 18:29 | michaelcubox | ok. lets hope with the cubox pro the development will go further |
| 18:30 | michaelcubox | i'm not a kernel expert, i did administration of debian machines and know debian by brain now, but i'm studiyng right now. |
| 18:37 | rabeeh | hi michaelcubox |
| 18:37 | michaelcubox | hi |
| 18:38 | rabeeh | shesselba works with 3.7 DT |
| 18:39 | michaelcubox | does he have a github or something we can use for distro? |
| 18:39 | michaelcubox | for 3.7 to clone i mean |
| 18:42 | rabeeh | i searched the log but didn't find. |
| 18:42 | rabeeh | shesselba lives at nights; he will probably be online in few hours. |
| 18:43 | michaelcubox | ok, which client do u use for irc? at the moment using webclient but it doesnt show if people are online or not. maybe pidgin? |
| 18:44 | lubiana | michaelcubox: i used irssi for a long time |
| 18:44 | lubiana | its a commandline client |
| 18:46 | Kiranos | quassel |
| 18:47 | Kiranos | like screen+irssi but graphical interface |
| 18:47 | Kiranos | just place a core/daemon on a server and stay connected 24/7 |
| 18:47 | Kiranos | just use the quasselclient to access your core |
| 18:48 | Kiranos | I even think there is a statically linked core binary so just drop and run basicly :) |
| 18:49 | michaelcubox | back with xchat now ;-) |
| 18:50 | michaelcubox | @rabeeh: xserver-xorg-core is not available in version 1.13 in debian armhf |
| 18:51 | michaelcubox | version 1.12 is in wheezy, in experimental there is no version at the moment |
| 18:51 | rabeeh | ok |
| 18:51 | rabeeh | can u rebuild and send me tarballs? |
| 18:51 | rabeeh | seems a downgrade is a safe path here |
| 18:55 | michaelcubox | i would like to do this, but i dont have that much experience in package building and also would have to learning. Just compiled a 3.6.9 for the debian rootfs by the way |
| 18:56 | michaelcubox | learn student stuff |
| 18:57 | michaelcubox | for next version of debian armhf build it will contain a headless and a desktop kernel. maybe that way we can get all satisfied. |
| 18:58 | michaelcubox | after april i've got time for this stuff |
| 18:59 | rabeeh | i'll rebuild those then |
| 18:59 | rabeeh | ubuntu 12.10 packages seems ahead of debian |
| 19:04 | michaelcubox | yeah, they always grab the latest from experimental and then do their own work with these. Every 6 months they do the same and upgrade was always a pain. Debian is designed for server, not for dekstop, laso i do run it at desktop |
| 19:05 | michaelcubox | a way to go is using sid when wheezy is released. newer packages, but with stability of stable release |
| 20:24 | Alphadog | someone tried to compile irssi-0.8.15 on geexbox ? |
| 20:25 | Alphadog | when running configure it says it cant find a compiler (c cc or gcc) |
| 20:26 | Alphadog | opkg install gcc-final said it is installed but cant find a gcc in /usr/bin |
| 20:35 | tomlohave | there is no compiler installed |
| 20:35 | tomlohave | all packages are cross-compiled |
| 21:15 | NotRelevant | http://openblocks.plathome.com/products/ax3/ |
| 21:19 | NotRelevant | http://debian.2.n7.nabble.com/Re-Debconf-discuss-OpenBlocks-AX3-A6-td1673067.html |
| 22:05 | michaelcubox | is anybody else epxeriencing that the solidrun server is very slow? |
| 22:06 | michaelcubox | just using the cubox installer to test something and download speed is average 78793 bytes/s |
| 22:21 | niteye | i dont know yet, i got my cubox yesterday |
| 23:21 | michaelcubox | wishing you luck ;-) |
| 23:55 | shesselba | michaelcubox: 3.7 contains basic cubox support, no audio/video/decoding |
| 23:55 | shesselba | I guess audio will be next as it is already there but needs clean up |
| 23:55 | michaelcubox | no problem for me as i use cubox as headless server ;-) |
| 23:55 | shesselba | 2d video should come soon (_rmk_ ?) |
| 23:56 | shesselba | decoding will take a while |
| 23:56 | michaelcubox | ok. so is there a github or something where i can grab it from and compile? |
| 23:56 | shesselba | git.kernel.org? |
| 23:57 | michaelcubox | well, i tried to compile but it did stuck at some points. |
| 23:57 | michaelcubox | i tried to get a similar setup to the 3.6.9 one with menuconfig but it did always complain about a machine thing not set |
| 23:57 | shesselba | try 'make dove_defconfig' |
| 23:58 | neofob | michaelcubox: where do you get your 3.6 linux repo? |
| 23:58 | shesselba | well you need to append cubox dtb |
| 23:58 | shesselba | u-boot on your cubox does not support DT |
| 23:58 | michaelcubox | where do i append cubox dtb? in dove_defconfig? |
| 23:58 | michaelcubox | @neofob: from rabeehs github |