=== dyfet_ is now known as dyfet
=== cbrake_away is now known as cbrake
jordanuclaHi all, is there somewhere that lists the differences between ubuntu and ubuntu-arm? Going to build a little home NFS/music/X10/torrent server so thinking to use ARM, but wondering if it will be powerful enough for the basic things I need.16:30
Martynjordanucla : Just about all the packages in main, and most of the packages in universe are there16:31
Martynas for "powerful" .. umm .. it depends on the processor?16:32
MartynIf you're running a Shivaplug, it's going to be 'meh'16:32
Martynif you're running a babbage board, it's going to be 'meh+'16:32
jordanuclawould you say Arm processors are usually equivalent to x86 in terms of model?16:32
jordanuclai.e. 800mhz arm == 800mhz x8616:32
jordanuclaor not that simple?16:32
Martynthat question makes no sense16:32
MartynRISC vs CISC16:32
jordanuclaya, poorly phrased16:33
Martynentire architectural differences16:33
MartynL1 cache in an ARM is split between data and instruction caches16:33
MartynL2 caches are tiny (256k, or smaller in some cases)16:33
Martynthey are peppy16:33
ograthe have arms ... the others dont :)16:33
Martynand have far more MIPS per watt for lower $$16:33
Martynogra : We got two PBX-A9's in16:34
MartynThey run at 70Mhz, but have dual (and triple) core action16:34
jordanuclawould a faster one of the ARMs be sufficient do play high quality video, flash, stuff like that?16:34
Martynso I finally have real hardware to work with16:34
Martynjordanucla : There is no flash player for ARM yet16:34
jordanuclaah, still porting16:34
* ogra just had to send his hw away :/16:34
MartynAdobe is working on the port16:34
jordanuclaya, I know they started16:34
Martynogra : Which hardware?16:34
Martynawww .. b2's?16:35
ograi'll get it back though16:35
jordanuclawhat about video and audio codecs? are they pretty widely ported?16:35
Martynor b1's?16:35
Martynjordanucla : all the codecs are ported16:35
Martynjordanucla: THAT works :)16:35
Martynand I can easily decode 1080p on a Cortex-A8 based system16:35
ograthe prob are most of the time the graphics drivers16:35
jordanuclaone last question --16:35
MartynIn fact, most MID's are based on that chip now16:35
ogranot the codecs16:35
Martynogra : Yeah, I'm getting really tired of the crappy implementation for the PrimeCell video driver16:36
ograMartyn, well, there is no opensource powerVR driver16:36
MartynI'm now working with Kevin W. at ARM to make a better one16:36
ogracool !16:36
jordanuclawith ARM is most hardware/boards/etc, mostly plug and play like x86. Or is it more like the older systems where things are treated as embedded and every ARM system can be very different?16:36
Martynogra :PowerVR will be next on that list16:36
jordanuclaor will it usually mostly just work with ubuntu-arm16:36
Martynthe latter jordan, unfortunately16:36
Martynwe're working on that16:37
ograjordanucla, every arm system is largerly different16:37
Martyn<-- is on the UEFI team now16:37
MartynWhich will bring a consistent bootloader to ARM16:37
jordanuclais there a list of recommended/supported configurations?16:37
ograbut the images we build in ubuntu are close to plug and play for the specific device we build it for16:37
jordanuclaI see16:37
jordanuclathanks for the hard work guys16:37
Martynogra : And even devices you don't build it for.  The armv5 buildroot has sped up my port to PBX16:37
ogralike the imx51 image we released functions similar to any x86 image you are used to with only minor differences16:38
Martynogra : I'll need help setting up a new build though, so that I can start compiling arm v716:38
jordanucladoes the webpage have a list of supported/recommended devices?16:38
Martyn(and v6)16:38
ograi'm just starting project rootstock, which is the former "build-arm-rootfs" ... you will be able to checkout the bzr branch and merge fixes16:39
Martynogra: Oh, and happy day .. beagleboard has seen the light, and will start adding SMSC ethernet to the beagleboard16:39
Martynogra : nod16:39
ograjordanucla, well, the out of the box thing will only be imx51 from karmic on ... and if we get the code in time another SoC i cant talk about yet16:40
MartynBut it wil be a 'D' version rather than a variant of the 'C's16:40
ograso if you want something that works out of the box try to get your hands on a pegatron laptop or some other imx51 based device16:40
Martynogra : N E C :)16:40
ograis there a spec for "D" already ?16:40
Martynogra : It's coming together16:41
Martynogra : I'm hoping to get my hands on some NEC silicon16:41
Martynthey have an a9 w/ dual core16:41
Martynbut getting my hands on the hardware is going to take dicey negotiations by the business team16:41
ograwell, for the "other" board i was mentioning above we might even get free GL drivers that work :)16:41
MartynOh, THAT board!16:41
* Martyn is so jealous16:42
ograbut its still not clear we get all code drops in time for karmic16:42
Martynit's not an A9 though :)16:42
ograa6 with 7 extensions16:42
MartynI'm not even allowed to look at non v7 anymore :(16:42
ograso call it a6+ :)16:42
Martynyep, I know it.16:42
ograit is v7 compatible16:42
MartynIt's kind of a weird mishmash core16:42
Martynno, it's v6 with some v7 instructions16:43
ograenough to make it speedy and use a6+vfp16:43
ogranot so sure about NEON though16:43
Martynthe arm-cc compiler will do it, and eventually codesourcery too .. but how long will it be before we see support for that mish-mashy processor in gcc-main?16:43
ograwe will16:43
Martynmuch less real NEON in gcc-main?16:43
Martynwhich I'm waiting for16:44
ograsimply because we build images for that board16:44
Martynnot to mention tons of corrections which are needed in glibc16:44
Martynbecause memcpy/memmove are horrible right now16:44
MartynI did some performance testing, and things are much slower than they should be, and could be16:44
* ogra heard different things ... but i dont have the HW yet16:45
Martynogra : Once I have a kernel booted on PBX1 and PBX2, I'll give you a remote shell in to look16:45
Martynit's ... interesting in there16:45
Martynalthough I'll be using our v5 jaunty rootfs16:45
Martynlool will have to show me how to do a full ubuntu build with v6 (or v7) optimizations16:46
MartynAs it is, I have today off :)16:49
MartynTomorrow is American Indepedence day, so hopefully I'll have some fun16:49
broonieDoes anyone have any info on how the process of upstreaming the ARM kernel changes you guys are carriying is progressing?16:49
Martynbroonie : which changes?16:49
Martyn(There is a patchwork of patches)16:50
Martynmost things end up in the arm linux org git tree16:51
ograbroonie, amitk would know ... but i think a good bunch of the imx51 stuff was rejected last time16:52
Martynhowever freescale is resubmitting a number of the patches16:54
broonieMartyn: drivers, architecture...16:55
broonieogra: Hrm, I hadn't noticed any of it going past TBH. Thanks.16:56
ograwell, he tried to get it in for jaunty ... not sure whats the current status for .30 and above16:56
Martynogra : mismash of things happening on the linux-arm.org git tree, but I haven't seen the patches merged yet16:57
ograas i said, rejected16:58
=== bjf_afk is now known as bjf
broonieThat's just the thumb stuff they're carrying, isn't it?16:58
Martynthere are also a bunch of Cortex-A9 patches that are about to hit.16:58
bjfbroonie, upstream want all the imx51 changes rewritten16:58
broonieWhich upstream?16:59
bjfbroonie, as ogra said16:59
broonieogra: Who/what do you mean when you say linux-arm?16:59
Martynlinux-arm.org, specifically Russel King17:00
broonielinux-arm.org isn't rmk.17:00
broonieI think...17:00
Martynall ARM inclusions into mainline have to pass rmk's inspection17:00
broonieYeah, that's arm ltd.17:00
* Martyn has a lot of rewrites to do as well, to support SMP on arm properly17:01
broonieThat's what I was querying - if it was linux-arm-kernel/rmk.17:01
bjfbroonie, i believe it is the penguintronix guys specifically17:01
broonieor if it was some other people who feed into them (eg, imx changes will need to coordinate with the pengutronix guys).17:01
broonieAh, makes sense.17:02
armin76Martyn: smp on arm? cortex a9?17:05
ograarmin76, yeah, he is showing off again with his ricer hardware *g*17:05
ograarmin76, 70MHz dual cores :)17:08
Martynarmin76 : Yep17:11
MartynHey, at least it's hardware, and not just FastModel or System C simulation17:12
Martyn(although the FastModel runs at about 125Mhz)17:12
Martynarmin76 : Still, getting everything working, and getting everything working _well_, are two different things17:15
Martynmy job is to get all the ACPI modes supported17:15
Martynand that's a big, tall order17:15
MartynRight now, we have S3.  That's it17:15
MartynSo you can sleep, and you can wake.  Woo.17:15
MartynThat's not a lot of power savings17:15
MartynI need to get cpuidle working well, get a couple better governors in there, etc.17:16
ian_brasilogra, project rootstock sounds cool17:32
ograthanks :)17:33
=== bjf is now known as bjf_afk
Shmeckanyone here by any chance?21:30
=== cbrake is now known as cbrake_away

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