/srv/irclogs.ubuntu.com/2010/03/02/#ubuntu-arm.txt

twbpersia: mainly I wanted an arm netbook for the battery life, and so I had an excuse to play with arm01:22
persiaWould you accept a 5" Netbook with a 7-8 hour battery life?01:23
twbBut I've basically been sold on the new AMD64 Atom that came out in December, which has 8 to 10 hours.01:23
persiaWhat's the mass of a shipping product?  At least for me, grams matter more than architecture.01:24
twbpersia: my netbook's primary job is to connect to an external monitor and keyboard, and run Emacs and SSH to connect to "real" computers (i.e. stuff in the racks) for sysadmin work and for writing code.01:26
twbIt's secondary role is to let me do the same stuff while I'm not at my desk, e.g. when I'm at a friend's place or on a bus.01:27
persiaWell, if you're willing to run Lucid, you can use a USB dock with an arbitrary device to take care of the primary job.01:28
persiaThe secondary role becomes more interesting: and then it's a matter of size/mass/time01:29
twbIt'll run Debian, not Ubuntu.  Sorry.01:30
persiaWell, I hope to find time to get displaylink working for squeeze, but I'm not sure I'll finish testing prior to freeze.  It will definitely be in squeeze+1.01:31
persiaBut then this isn't the right channel anymore :)01:31
twbYeah, I figured I'd just stick my head in to hear what you had to say re. the netwalker01:33
persiaI'm happy with mine, so I tell people who want ARM netbooks that there is one currently in retail and they should go buy it.01:34
persiaI'm not convinced that lots of others will appear in retail if the first one doesn't sell so well.01:34
twbWhat was it retailing going?01:35
twbWhat was it retailing *for?01:35
persiaEPARSE01:35
twb(Edited out the wrong word.)01:35
persiaDepends on where you get it.  Somthing like 500 AUD01:36
persia+shipping / etc.01:36
twbBlergh.  If it was AUD300 I'd be more interested.01:37
DanaGwait, arm netboook where? the touchbook?01:37
persiaYeah well.  This is why the rest of the world has to wait a couple years for Japanese electronics :)01:37
persiaDanaG: The NetWalker01:37
twbDanaG: we're talking about the Sharp Netwalker01:37
twbDanaG: http://en.wikipedia.org/wiki/Netwalker01:38
DanaGah.01:38
persiaNice little box.  I find the positioning of the "A" key a bit frustrating, but otherwise very usable.01:38
persiaSturdy too: I've watched it bounce down a flight of concrete&steel stairs, and dropped it from ~3.5 meters a few times onto hardwood.  I don't recommend doing this, but mine still works.01:39
DanaGhmm, does it do mobile broadband?01:40
DanaG(not that I have a data plan...)01:40
persiaDanaG: It's advertised as doing so with a USB dongle.  Having split my case open, I *think* there's enough space to wedge one inside, but it's a bit of a hack.01:40
persia(plus you'd be down to one USB port)01:41
DanaGbah, I'd want mini-pcie.... and you could use it for gps, while you were at it anyway.01:41
persiaFinding PCI or PCIe on ARM boards is *hard*.  I have a collection of ARM devices, but only one of them has PCI, and I'm not convinced it isn't a hack like PCI-over-USB on the boards.01:42
persiaThe bus usually isn't there at all.01:42
twbYou should be able to work it out by checking if the chipsets on the board implement pcie at all01:43
persiaI'm pretty sure the i.MX51 series doesn't.01:44
twbYeah01:44
DanaGErm, most "Mini-PCIe" WWAN devices are really USB-based.01:45
DanaGAnd most "Mini-PCIe" SSDs abuse mini-pcie pins for SATA or PATA.01:45
persiaSure, but if the bus isn't present, you can't expect the pinout to be available to be abused.01:46
persiaAt least for the Eee hacks, stripping the case from a USB dongle and sticking it internal seemed a popular solution.01:46
twbIs mini-pcie that crufty mishmash of pcie and cardbus?01:47
persiatwb: Yes.01:47
twbRighto01:47
persiaNot quite as confusing as XpressCard, but close :)01:47
DanaGWrongo.01:49
DanaGMini-PCIe is the internal slot -- equivalent of mini-PCI.01:49
persiaPlease share?01:49
DanaGExpressCard is the PCIe equivalent of CardBus.01:49
DanaGBoth mini-PCIe and ExpressCard happen to offer both USB and PCIe interfaces in the standard, though.01:49
twbSorry, I got confused.01:50
persiaOK.  So mini-PCIe is USB+PCIe and ExpressCard is CardBus+PCIe+USB ?01:50
DanaGnope, ExpressCard is just hotpluggable PCIe+USB in a cardbus-ish shape.01:50
Martynyep01:50
Martynconvenient too01:51
persiaSo the old CardBus ATA hacks don't work?01:51
Martynalthough the cardbus-like connector puts some serious constraints on the types of peripherals you can create01:51
DanaGAnd ExpressCard 54mm actually offers LESS space than cardbus.01:51
persiaMartyn: Bandwidth excepted, how?01:51
Martynpersia : Mini-PCIe can handle larger current draw01:52
persiaAha!  The EeePC has a *special* *custom* miniPCIe that is PCIe+USB+SATA01:52
Martynit's designed for it01:52
twbpersia: which eee is that?  I'll avoid it.01:52
MartynWhich pins do they overload to get SATA?01:52
persiaMartyn: That makes sense.01:52
persiahttp://beta.ivancover.com/wiki/index.php/Eee_PC_Research is cited in wikipedia as the reference for the odd custom port.01:53
DanaGAnd a bummer with HP: they whitelist stuff, so you can't use the wwan usb-only slot for anything but HP wwan module.01:53
Martynmust be pins 51,49,47,45,43,41,39,3701:54
Martynthose are "reserved"01:54
persiaDanaG: But seriously, you shouldn't worry about PCIe if looking for an ARM netbook.  It's unlikely to be implemented in most solutions (I would expect PCIe to only be exposed for "server" type devices (which might include plug-servers and NAS boxes).01:55
Martynpersia : *cough*  *COUGH*01:55
* Martyn can't say .. but *COOOOOOOUUUGH*01:55
Martyn(in fact, most people implementing cortex-A9's are implementing Synopsys' PCIe)01:56
persiaMartyn: Really?  That's very encouraging.01:56
DanaGI saw somebody on #radeon talking about sticking a Radeon on Marvell ARM thingy.01:56
MartynDanaG : most manufacturers of multicore ARM solutions are looking at PowerVR01:57
Martynalthough nVidia is getting hot 'n heavy with Tegra201:57
persiaDoes PowerVR have open-source drivers yet?01:57
DanaGGood luck getting open drivers for that, though.  =þ01:57
MartynDanaG : Open drivers now exist01:58
persiaOh excellent!01:58
MartynDanaG : They went through the trouble of creating an ABI + shim just like nVidia01:58
Martynso they can publish a binary driver, and open source (linux) solutions only have to implement the shim01:58
DanaGah, though, that's not entirely open.01:59
MartynSure, but open drivers DO exist, just not optimized, nor with great 3D accel01:59
DanaGrandom link: http://www.techeye.net/software/amd-and-nvidia-bitchfight-over-open-source-support01:59
persiaA shim solution is sufficient to start.01:59
DanaGah.01:59
persiaA parallel to the various more-open driver can wait.01:59
DanaGMy criterion for "good" 3D is "can run compiz".02:00
DanaG02:00
Martynhell, there are some OPEN SOURCE drivers that can't run compiz02:00
Martynfor well supported cards (points at nouveau)02:00
persiaLots of them.02:00
persianouveau can run compiz for some cards.02:01
Martynin fact, I can't think of any that can support the full compiz requirements02:01
Martynclosed source shims do so pretty well02:01
persiaThey have their own problems02:01
DanaGcheck the last comment on that thread... that's my post.02:01
Martyn<-- is _not_ biased towards open source.02:01
* persia has a shim-melted card02:01
DanaGWhat I call bad: when your binary drivers fail to do anything but segfault the X server -- and have been broken in that way for 2 years.02:02
persiaBut shims are sufficiently free to ship, which is the key bit.02:02
persiaDanaG: That's usually a case of full binary drivers, rather than shims.02:02
DanaGAnd they've "updated it to support new x servers"  -- but I say, s/support/segfault/02:03
DanaG02:03
DanaGanyway, how well does the SGX in beagleboard work?02:03
DanaGOh, and OpenGL ES (OpenGL on bare console?) seems like 'black voodoo magic' to me.02:04
twbGive me a Matrox GPU any day :-)02:04
MartynDanaG : That's a pretty good description :)02:04
MartynOpenGL ES uses the framebuffer as the base canvas02:04
Martynit's not THAT much voodoo02:04
Martynbut it's hard to grok02:04
=== bjf is now known as bjf-afk
persiaStevenK: So, I was just reviewing the actions from last week's meeting, and it appears you and I were supposed to strip armel down to three flavours : *-netbook and server.  Did you already do that?  Can I help somehow?09:30
Guest62340Hi09:55
persiahey09:56
Guest62340 I have installed Mer on my N81009:57
persiaGuest62340: OK.09:59
StevenKpersia: I didn't, no ...10:02
persiaStevenK: OK.  Can I help?10:02
* StevenK peers at antimony10:02
ograhow are netbook and server *three* flavours ?10:03
persiaogra: *-netbook10:03
persiakubuntu-netbook and ubuntu-netbook10:03
ograah10:03
StevenKWhich are currently there.10:03
* ogra still sees no point in kubuntu building as long as kde isnt installable10:03
StevenKpersia: Right, I think it's done, if you want to check10:04
persiaRight.  We want to not have ubuntu-desktop, kubuntu-desktop, xubuntu-desktop, and ubuntustudio-*10:04
persiaStevenK: I don't know how to check, but if you think it's done, I'll trust you.10:04
StevenKpersia: Dig through cdimage.u.c and see if armel images appear where they shouldn't.10:05
persiaThat I can do :)10:05
StevenKOr if they don't appear where they should10:06
persiaI don't see any ubuntu-server images at all :(10:08
persiaOh, just build failures.  Nevermind.10:08
StevenKHow do they fail?10:09
persiaDIdn't check that.10:10
persiaBut it's for all architectures for the 0301 build10:10
persiakubuntu-desktop is current on armel10:10
persia(and shouldn't be)10:11
persiaerr, kubuntu-desktop-live (sorry)10:11
persiaIn summary, kubuntu-desktop-live needs to go, and ubuntu-server-alternate needs love.10:13
StevenKI can see kubuntu desktop in the cronscripts10:17
StevenKNCommander: ^10:17
persiaWhat do you seek?10:18
persiaI can produce logs from #kubuntu-devel that agree it doesn't belong there10:18
persiahttp://irclogs.ubuntu.com/2010/02/23/#kubuntu-devel.txt 14:09 through 14:1410:19
ogralool,      if [ "$DIST" = "jaunty" ];then10:35
ogra         VMCPU=""10:35
ogra     fi10:35
persiaUm, why?10:36
persiaShouldn't it detect current compilation options or something?10:36
StevenKpersia: Right, fixing crontab, then10:42
persiaStevenK: Thanks.10:42
StevenKpersia: Fixed, and old images purged10:47
persiaStevenK: Excellent!  Now we have good news to report *and* more space on cdimage for the next hungry flavour.10:48
StevenKpersia: Build log for swerver?10:49
* persia gets confused10:51
persiaAha!  The issue was that the images were *currently building* when I last looked.  They were successful and are now published.10:52
StevenKRight, so nothing more to do10:53
StevenKWe win, are manly men, etc, etc10:53
* persia waves the torch of victory and ignores the sounds of looting and pillaging10:57
ograsooo ...10:58
ograwhat do you guys think about adding a date -s "<date of image build>" to casper for armel ?10:59
persiaWhy would such a change be architecture-specific?11:00
persiaI like the idea, but I don't see the point of making it arch-specific at all.11:00
ograbecause it bugs us most11:00
ograbut yeah, can be generic11:00
persiaNo, it really doesn't.11:00
persiaIt bugs me *lots* but mostly for i386 images.11:00
ograwell, PC BIOS batteies are less often broken than having boards that dont set the time on first boot11:01
persiaOh, you're not talking about the thing that irritates me at all :)11:01
persiaBut anyway, yeah, I don't see any issues with it, and don't think it ought be arch-specific.11:02
persiaWe know it's at *least* that date.11:02
ograright11:03
loologra: Are you pinging me about the VMCPU thing?11:03
ogralool, yeah, i did, but i gathered its not needed if we use the netinst kernel for all releases11:04
loologra: Note that I only *flipped* the logic to deal with karmic +11:04
loolSee r30.1.111:04
ograi dropped all release specific kernel stuff now and default to not use devtmpfs in rootstock11:04
loolhttp://paste.ubuntu.com/386913/11:04
loolI changed the logic which only worked for jaunty and karmic to work for jaunty and any release11:05
ograthat shpould make it work with the lucid kernel everywhere11:05
loolBut we should simply set Cortex-A8 ALL the time11:05
ograright, thats what i do now11:05
ograusing only the netinst kernel11:05
loolOnce the binaries are new-ed, we should have initrd support in that kernel as well -- I hope11:06
ograwell, doesnt bother me in rootstock atm :)11:07
ograits unlikely i'll use an initrd there11:07
ograthough i could generate one during build if --keepimage is set for people wanting to use one with their VM11:08
ogragrrr11:09
ograevolution still out of sync ... i cant do testbuilds11:09
persiaJust give up on the concept of email.  Adds more time to your day and you never have sync failures :)11:11
ograpersia, well, then someone needs to fix the metapackages :P11:11
persiaHrm?11:12
ograits not installable atjm11:12
ogra*atm11:12
persiaOh, right.  I misunderstoof.11:12
* ogra takes a break11:24
* StevenK throws evolution back to the wolv^Wbuildds12:05
twbAnd I thought *emacs* would stress a 1GHz arm...12:10
suihkulokkitwb: building vim takes longer than building emacs ;)12:13
twbEmacs is "only" eight megabytes and constantly swapping :P12:14
twbWhereas eclipse is more like eight HUNDRED12:14
jussi01hi all, http://www.ubuntu.com/products/whatisubuntu/arm says the iMX51 is supported, but will it run on an iMX31?12:19
ogranope12:20
jussi01aww...12:20
ograthe kernel on these images is specifically built for a certain board12:20
jussi01and hi ogra! :D12:20
ograhi :)12:20
jussi01ogra: right.12:21
ogra****************** Reminder mobile team meeting in #ubuntu-meeting in 3min ***********************12:57
twbWe aren't all on 132-character terminals, you know :P12:58
ogralol, sorry12:58
ograi'll make sure next week its 80chars :)12:58
twb80 less your nick, etc.13:01
persiaImpossible to know how much our clients prefix, etc.13:02
twbBut actually my framebuffer is something like 300x12013:02
persiaJst dropping the *s would probably work13:02
ograah, come on, get a wider screen :) its not like i have a long nick :)13:02
twbAren't the **s a dumb way to use /notice ?13:02
ograyeah, i'm lazy13:02
twbHmm, didn't work: 00:03 /notice #ubuntu-arm Yow!  Film at 11!13:03
twbOops, I'd better wrap up before security come to hassle me again13:03
* persia doesn't tend to like arbitrary /notice13:05
persia(and would go fiddle flags if people started using it much)13:05
twbpersia: using it judiciously is better than it being totally unused except by the freenode devs14:10
twb(IMO)14:10
persiaI guess.  I just don't really like it.14:10
=== bjf-afk is now known as bjf
=== InfecteD is now known as Guest77043
|nfecteDanyone know how i can limit how much RAM the beagleboard has avalible... it fails horribly right now when i try to compile larger programs21:31
persia|nfecteD: You mean limit how much memory is available to the compilation process?21:32
|nfecteDyeah21:33
persiaThe solution I typically use in memory limited situations is swap, but swap-over-USB gets painful quick.21:33
|nfecteDI know there was a file or something in angstrom where i could change a value from unlimited to something else and it worked :/21:34
persiaYou can try playing with `ulimit -v` but it may just make the compiler crash more quickly (depending on the operation)21:34
|nfecteDUSB swap it is then21:34
|nfecteDi need a bigger SD card21:34
persiaswap-on-sd is a recipe for buying new SD cards :)21:35
|nfecteDyeah21:35
|nfecteDsame goes for USB sticks i guess21:35
persiaBut not for old-style USB rotary disks :)21:37
|nfecteDi only need to use "swapon /dev/sda" right?21:37
|nfecteD(after i format ofcourse)21:37
persiaSoemthing like that.  I always put stuff in /etc/fstab and use swapon -a when it's available.21:37
persiaWhere "format" means "run mkswap", yes :)21:38
|nfecteDgood thing you know what i mean21:38
* |nfecteD throws 2gigs of swap at the compiler21:39
|nfecteDfill that up!21:39
|nfecteD(bastard...)21:39
persiaBuild boost :)21:39
persia(no not really : boost is best built either 1) when really necessary or 2) when one wants to exercise high memory-pressue situations)21:40
|nfecteDif i could find my USB HDD i'd have 30GB swap!21:41
|nfecteDSUPER SWAP POWER!21:41
|nfecteDanyway... isn't there supposed to be some way to set the CPU frequenzy while im in ubuntu?21:41
dyfetyou mean like cpufrequtils?22:00
|nfecteDi guess22:00
dyfetI recall using those on debian...22:00
|nfecteDsince i have the ub3rl33t C4 version that can "safely" be clocked to 720mhz22:01
|nfecteDi might as well use it sometimes22:01
roxfano.o http://www.keil.com/ds5/22:02
|nfecteDgod damnit!¤!¤!¤22:04
* |nfecteD curses and swears at the snes9x sourcecode22:04
|nfecteDboo22:14
|nfecteDi don't have a cpufreq driver22:15
persia!ohmy22:27
ubot4Please remember that all Ubuntu IRC channels share the same attitude of providing friendly and polite interaction with all users of all ages and cultures. Basically, this means no foul language and no abuse towards others.22:27
* |nfecteD sits in the corner22:31
persiaheh :)22:33
persiaJust try to say #('$& instead or something :)22:34
persiaNot much to self-censor, but in addition to other stuff, it keeps our logs free from nanny-filter blocking.22:34
|nfecteDyeah22:36

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!