[03:56] that is odd [03:56] I ran sudo apt-get source linux-image-$(uname -r) [03:57] uname -r is 3.1.0-1282-omap4 [03:57] and it downloaded linux-ti-omap4_3.0.0-1205.10.tar.gz === XorA|gone is now known as XorA === Sv is now known as discopig [09:22] anyone up? [09:24] yes [09:25] good to know i'm not the only one. [09:25] well its 11:25am, even in France we are awake :_D [09:26] anyone (over in europe) know if it is possible to install a multilib arm system, such that both armel and armhf executables will work? [09:26] it looks like the default 12.04 system does not have this functionality out of the box. [09:27] mjrosenb: it is actually possible, but it requires special builds, arm libc can use different varients depending on the capabilities flags [09:27] mjrosenb: but in reality I think it lead to insanity [09:28] some really clever person in open moko days even got OABI on EABI loader working [09:28] XorA: It doesn't require special builds. [09:28] XorA: And it has nothing to do with capabilities. [09:28] needing to use an 11.10 install just so Ican have an armel system isn't very sane [09:29] mjrosenb: 'echo foreign-architecture armel > /etc/dpkg/dpkg.cfg.d/multiarch' [09:29] mjrosenb: And apt-get update. [09:29] mjrosenb: And then you can install foo:armel [09:29] mjrosenb: Like, say, libc6:armel, if that's what you need. [09:30] * mjrosenb tries it. [09:30] XorA: Caps can't do this at all, armel and armhf are ABI incompatible. [09:37] sorry was thinking of soft vs softfp [09:37] arm really needs to stop making up new ABIs! [09:39] libc6:armel seems to be installing! [09:40] hey cool, it is dying with a different error! [09:41] ugh... dependency conflicts [09:42] why do I have to run "apt-get install libstdc++6-4.6-dev:armel g++-4.6:armel gcc-4.6:armel cpp-4.6:armel binutils:armel", shouldn't debian be able to resolve everything as a dependency when I run apt-get install libstdc++6-dev:armel [09:43] * mjrosenb also wonders if gcc-4.6:armel is a cross-compiler to armel, or if it is a native-compiler for armel [09:44] update-alternatives: using /usr/lib/arm-linux-gnueabihf/mesa-egl/ld.so.conf to provide /etc/ld.so.conf.d/arm-linux-gnueabihf_EGL.conf (arm-linux-gnueabihf_egl_conf) in auto mode. [09:44] that looks like it could be bad [09:46] http://pastebin.mozilla.org/1870782 [09:46] ok, that looks *much* worse [09:48] http://pastebin.mozilla.org/1870783 [09:48] well, that's fun. [09:57] mjrosenb: gcc-4.6:armel should be a native armel compiler [09:58] phh: so if I install gcc-4.6:armel on an x86 machine, I'll get a compiler that I can't use? [09:58] 1.i doubt you can install it; 2.there could be the qemu stuff involved === doko_ is now known as doko === Guest98671 is now known as Stealthy`Office === Stealthy`Office is now known as Stealthy === tdc_ is now known as tdc === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [13:47] ndec: any plan to get DVFS upstream? [13:48] hehe [13:48] it will, eventually. [13:48] i'll take it as "we've no ETA" [13:48] that is a big task, and it requires developing a generic ARM framework. many people are working on that right now... [13:48] it will be there, when it's ready ;-) [13:48] ok [13:50] http://victorpalau.net/2012/10/19/nexus-7-raring-to-go-to-copenhaguen/ [13:51] :D [13:53] eheh [13:53] ogra_, hah nice, do you happen to know if that is with the tegra libs? [13:54] lilstevie, thats fully running GLES,yes [13:55] oh nice [13:55] good to know that trusted foundations doesn't break the libs [13:55] nux was fixed for tegra in that branch [13:55] the fixes will flow back into mainline nux [13:56] yeah, that wasn't really the biggest part that I was interested in :) [13:56] the biggest part I was interested in is that the nexus 7 bootloader (and kernel) have trusted foundations enabled [13:56] and the libs, well, don't [13:56] in android it matters [13:58] doesnt in ubuntu :) [13:58] and the nexus7 ids fully open [13:58] clearly :p [13:58] took me a day to get a full native installation working [13:59] yeah, well my main concern as I said was trusted foundations :p I have actually been holding off with the tf201 cause I haven't updated yet [13:59] asus enabled it on the jellybean update for us, involving a very fragmented experience [14:00] tf enabled kernels don't boot on non tf bootloaders and vice versa [14:07] ah, well, the nexus is just like the ac100 [14:07] not much difference apart from double the ram and double the cores [14:10] and the addition of a trustzone driver :p [14:10] point is at least the nexus is consistent [14:11] all n7s have it, not just those with the latest firmware update [14:12] i'm not sure it is used at all after you unlocked the bootloader [14:12] we definitely dont have any TF firmware in the image [14:12] and i dont think TF works without blob [14:12] it is in the kernel though [14:12] ogra_: so good ubuntu is on nexus7? [14:12] s/so/so how/ [14:12] that is besides the point if it is enabled in the bootloader, it needs to be enabled in the kernel :p [14:13] hrw, yeah [14:13] hrw, its fine, faster than on panda [14:13] ogra_, turn off CONFIG_TRUSTED_FOUNDATIONS and it will fail boot :p [14:13] ogra_: two cores more and faster gpu helps I think [14:14] I would love to see unity3d on the tf201 [18:53] hey, has anyone used oprofile on 12.04? [18:56] #linaro might [19:52] ok, slightly easier question [19:52] is oprofile included with 12.04? [19:56] I thought all the cool kids used perf these days [19:56] yes, but perf does not have jit integration. [19:56] ah [19:58] and I ended up building oprofile from source, since it looks like it is not in apt [19:58] http://pastebin.mozilla.org/1871282 [19:59] well, did you ask #linaro ? [19:59] ubuntu usually takes what they provide [20:03] mjrosenb: From what I can tell, precise omap4 kernel has it enabled. [20:11] GrueMaster: yes, the kernel has it (onieric did not) [20:11] GrueMaster: but the actual package itself does not have it [20:11] ogra_: I did, they told me fridays are bad, and gave me a mailing list to email [20:11] my email there bounced :( [20:14] Hmm. Not even in the x86 repo. [20:16] well, that isn't *too* annoying. === Ursinha is now known as Ursinha-afk [21:44] soo [21:44] this place will probably be buzzing in a few days [21:45] anyone in here planning to pick up the exynos chromebook? [22:02] * GrueMaster is tempted, but may have to wait a bit. [22:03] as soon as they get into BB i'm going to go play with one [22:03] for $250 it's highly tempting [22:12] SoulShadow: is that the one that samsung is making? === Shadows` is now known as SoulShadow [22:16] mjrosenb yes [22:16] yeah, if fry's has any next week, I am totally going to purchase one. [22:19] they're on amazon [22:22] yes, but i'm going to be spending the week in california, not at home, so having one delivered here does not do me any good. [22:31] ahah [22:42] the only thing really holding me off from buying one mjrosenb is support for other os's [22:42] since i don't have az whole ton of faith in chrome os [22:42] is there any a15 support in ubuntu? [22:42] SoulShadow: From what I understand, the bootloader is unlocked, so if you have a kernel (which should land in Ubuntu for 13.04), you can run Ubuntu. The rest of the code in the repo should just work. [22:42] so stuck with chrome os till next year [22:42] unless someone writes a kernel earlier than 13.04? [22:43] From what I understand the kernel source is available, but will be pushed mainstream in 3.8. [22:43] hm [22:43] well, someone should be able to compile 12.10 w/ 3.8 [22:44] i would imagine it all depends on how much real interest there is in this device [22:44] it's #1 on amazon in laptops [22:46] Well, considering kernel 3.7 isn't even out yet, 3.8 should fall inline with the 13.04 timeline. And I'm sure it will be in backports for 12.04 & 12.10. [22:47] googling found me some reference of 3.7 a15 support [22:47] looks like xen support was merged in 3.7 [22:47] for a15 [22:48] so that would mean a15 is already supported, right? [22:48] Yea, it has been trickling in. There should be more SOC's coming soon. [22:48] so there's a chance that 12.10 w/ 3.7 might work [22:48] I'd say half backed. Core support is in the kernel, but may need tweeking once real systems come out. [22:49] Kind of like Intel's Haswell video support. [22:49] yeah [22:49] imean, as long as it works [22:49] it will be supported [22:50] But the fact that it is in the kernel means it should work (somewhat) when hardware arrives. This is actually very cool for Linux. [22:51] this is possibly one of the first truly mainstream devices that linux will be really focused on with [22:51] I wouldn't say that. Look at android. [22:52] i don't consider android to truly be 'linux' [22:52] in that sense [22:53] Well, until the hardware settles, there will be limited support in the mainstream kernel. Expect distro's to have patch sets on top of that to cleanup and optimize functionality. [22:54] i mean [22:54] chrome os can't be THAT bad as a daily driver [22:55] If all you want to do is browse the web, sure. But I don't think Chromeos is a GP OS. [22:56] well at the same time the hardware isn't exactly on the levels of a normal laptop [22:58] Define "Normal". I did a lot of Banshee debugging while flying to UDS Orlando last year on an AC100. [22:58] normal as being an intel based laptop [22:59] If you wanted normal, you wouldn't be in this channel. :P [22:59] i'm typing from an inspiron 1520 with an ssd lol [23:00] Besides, the AC100 took up half the space of my Acer Netbook, and had 2x the battery life. [23:00] I'm typing from a Windows 7 based T60 (new job - sucks). [23:00] doesn't the t60 have the high res 4:3 display [23:01] Not sure. Oops. T61 (my bad). It is a Centrino based system. My netbook can run circles around it. [23:02] lol [23:02] i'm curious about true battery life of the notebook [23:02] I actually run a remote desktop on an Ubuntu platform for doing real work. Just IT won't support it on the wifi network. [23:02] they say 6.5 hours [23:02] but [23:02] that is very low [23:02] I was using a T61 until just a few months ago, I'm not sure how anything could run circles around it... [23:02] Hard to say. [23:02] Unless you just meant the battery life. [23:02] For a 3-cell, that is about right. [23:03] my desktop runs circles around it :P [23:03] it's a 30whr it seems [23:03] but, tablets with 30whr batteries are getting much higher battery life [23:03] Try running Windows 7. [23:03] I gamed in Win7 on it. [23:03] and this is a tablet chip in a netbook [23:03] i feel like it's very conservative [23:03] or at least, i hope it is [23:04] infinity: With Intel Integrated Graphics? Really? [23:04] he's probably got an upgraded one [23:04] GrueMaster: No, mine had the nvidia discrete. [23:04] We're talking bottom barrel. [23:04] not a bare minimum model [23:04] And 8G of RAM. And such. [23:04] It was no slouch. [23:04] Well, there you go. [23:05] The CPU in the T61, at least on the top end, was faster than my current WhateverBridge. [23:05] Mine has 2G, Intel Graphics, and a slow 250G Sata drive. [23:05] Well, even at those specs, it's a heck of a lot faster than an ac100. Just sayin'. [23:05] But yes, the batter life sucks. :P [23:05] My netbook at least has a 120G SSD and 2G. [23:06] The only issue I had with the AC100 was ram and storage. [23:06] [19:04:54] The CPU in the T61, at least on the top end, was faster than my current WhateverBridge. [23:06] no. [23:06] i don't care what cpu it is, it's not faster than sandy/ivy [23:06] GrueMaster: hey, you seemed to know something about oprofile last night, you up for helping me figure out why it isn't working? [23:06] My battery is so old, it barely survives walking to the lab from my cube. [23:07] mjrosenb: I really don't, other than I can look on my pandas at home to see if it is supprted. [23:08] I haven't used it. [23:08] GrueMaster: ahh. oh well. [23:08] SoulShadow: For multi-threaded builds, my old machine was faster, yes. This Bridgy thingee shines for single-threaded tasks when it gets to overclock itself and go all turbo. [23:08] infinity: uhmmm [23:08] mjrosenb: Sorry. [23:08] what is your processor [23:08] because sandy and ivy are faster period [23:08] i7-2640M [23:08] no questions asked [23:08] k, you have a dual core w/ HT [23:08] sandy bridge [23:08] Yes, I know what I have. [23:08] what was the other one [23:09] The other one is on another continent right now. [23:09] do you know what it was? [23:09] x9100? [23:09] Don't recall. [23:09] I just know what "faster" means. [23:09] And it's not synthetic benchmarks or brochures. [23:10] well i'm TELLING you that the i7 is faster period, there's another bottleneck somewhere you're experiencing [23:10] Probably io bound. [23:10] Well, thank god you're TELLING me that, now I know. [23:10] UPPER CASE makes everything on the internet MORE TRUE. [23:10] you're telling me that a c2d is faster than an i7 [23:10] and that's a blatantly false statement [23:11] Or if the other system was a Core2Quad. [23:11] i don't believe t61 supported that [23:12] Not sure. Should be socket compatible though. [23:12] yeah the fsb difference makes it not post [23:12] the quads are 1066 [23:12] SoulShadow: I could imagine an incredibly high end core2quad being faster than a very low end corei7, but the fact that it is *bridge rather than nehalem makes it less beliveable. [23:12] mjrosenb: but it's not a quad [23:12] SoulShadow: Not on the Cor2 series. [23:12] GrueMaster: q9000 is 1066 [23:13] http://ark.intel.com/products/40480/Intel-Core2-Quad-Processor-Q9000-6M-Cache-2_00-GHz-1066-MHz-FSB [23:13] the t61 was 800mhz [23:13] Wikipedia claims it was a T9600. [23:13] http://ark.intel.com/products/33918/Intel-Core2-Duo-Processor-T9500-6M-Cache-2_60-GHz-800-MHz-FSB fastest proc in a t61 [23:13] (Which was a 1066 MHz FSB, FWIW) [23:13] because it's 800mhz [23:14] when i googled everything said 800mhz, infinity [23:14] so...that's odd [23:14] Meh, maybe it was the 9500. Remember the "other side of the planet" thing. [23:15] yeah, the t9500 is not even close to being as fast as an i7-2640m [23:15] at all [23:15] I was going to say the same thing, since I have both Core2Duo & Core2Quad systems at home. [23:15] Because you bench them together all the time? [23:15] um [23:16] because the c2d is really old, and sandy bridge is an evolution on that, with much higher clocks and a newer architecture with more cache and hyperthreading, therefore giving you higher overall performance [23:16] * GrueMaster wonders how the ubuntu-arm channel became a place for debating Intel spec's. [23:16] because he said that the i7 was slower than a c2d [23:16] and i almost spit my pop out [23:16] SoulShadow: Everything you said there ignored everything we know about CPU design. [23:17] SoulShadow: "Much higher clocks" imply "deeper pipelines" imply "CPUs at the same clock as the previous generation will be slower". [23:17] http://www.anandtech.com/bench/Product/54?vs=289 [23:17] SoulShadow: And my i7 is only 200MHz higher clock than my C2d. [23:17] highest end c2d (desktop) [23:17] um [23:17] vs lowest end i3 [23:17] sandy bridge [23:17] LOWEST END I3 [23:18] * GrueMaster wanders off from the SHOUTING MATCH. [23:18] your i7 turbos [23:18] you can't not take that into account [23:18] because it should almost always run at turbo clocks [23:18] SoulShadow: I kinda pointed out that my i7 doesn't turbo. [23:18] what [23:19] SoulShadow: Because that's not how multi-threaded loads work. [23:19] it still does turbo [23:19] just not as much [23:19] Infrequently. Not enough to make much of a blip. [23:19] o.O [23:19] do you have heat issues? [23:20] it should run at max (dual core) turbo during load [23:20] SoulShadow: It's a laptop, of course it has heat issues. [23:20] not all laptops have heat issues.. [23:20] It's a Lenovo laptop, of course it has heat issues. [23:20] i guess a better question is [23:20] there we go [23:20] that's what i was about to ask [23:21] Anyhow, I'm more than through with you talking down to me like I'm a seven year old child who has no idea what a computer is. [23:21] hah. [23:22] anyway, back to the more channel-appropriate discussion of the chromebook [23:23] i wish they put more storage in it, 16GB is too small. [23:24] Add a 32G or 64G SD. [23:24] true, but it's not quite the same [23:25] Depends on usage model. [23:25] If you are playing a movie or music, they are plenty fast. Even for doc storage. [23:25] It's when you are doing heavy compiles that it really hurts. [23:26] Of course, I have never liked the idea of using a laptop for heavy loads. They cost too much and weigh too much to handle the equivalent load of a remote desktop system. [23:27] does it have a media and music player? [23:27] lol [23:28] Not sure if chromeos does. Ubuntu does. [23:28] But you should be able to install Ubuntu into a chroot and run with it, worst case. [23:29] true, but yeah, a chromebook isn't for compiling [23:29] Neither is an AC100. But it can do it. [23:30] i wouldn't try it [23:30] what's the a100 run? [23:30] tegra 2? [23:30] yea. [23:30] ew [23:30] lol [23:31] With 512M ram and 8G eMMC. [23:31] do we have any info on the performance of a15 cores? [23:31] Like I said, for Ubuntu Desktop the only real limiter was the memory. [23:31] SoulShadow: I've heard a15 is much much better than a9. [23:32] it's faster sure, but we have no real benchmarks [23:33] the DDR3 in the chromebook should provide some nice performance [23:35] Again, it's all relative. From what I can see the sticking point will be with storage, but any linux distro can live comfortably in 4G or less. The rest is user storage, which can be supplanted with SD. [23:35] 2G memory on an arm platform is like 8G on an Intel platform. [23:35] does mali have good linux support [23:36] (note that is a usage comparison). [23:36] Mali is the GPU, right? [23:36] GrueMaster: but attempting to do dev work with 4G is a mit constraining [23:36] or eve debug [23:36] GrueMaster: yes, Mali is the gpu. [23:36] mali-t604 [23:36] quad core gpu [23:37] * mjrosenb will just be happy to have storage that isn't over a usb bus. [23:37] Like all Arm GPU's, linux support is binary only. [23:37] Depending on clock speeds I would expect peak performance north of the PowerVR SGX 543MP2, although I'm not sure if we'll see performance greater than the 543MP4. The Mali-T604 also brings expanded API support including DirectX 11 (feature level 9_3 though, not 11_0). [23:37] hm [23:38] Yea, cause Linux needs that DirectX support. [23:38] yeah, I was going to say... [23:38] i'm more amused by the fact they say 'dx11 supported' [23:38] but only supports 9.3 [23:38] how about opengl support? [23:39] supposedly supports 3.0 [23:39] They must be supporting Windows RT as well. [23:39] opengl es 3.0 [23:39] and opencl 1.1 [23:39] THAT is good. [23:40] what is openvg [23:42] it supports: OpenGL ES 1.1 -> 3.0, OpenVG 1.1, OpenCL 1.1, DX11, Renderscript [23:43] OPenVG - Vector Graphics Acceleration [23:47] still torn between nexus 7 and chromebook