ARM Publishes 64-bit "AArch64" Linux Kernel Support 90
An anonymous reader writes "ARM Holdings has made available Linux kernel support for AArch64, the ARMv8 64-bit architecture. No 64-bit ARMv8 hardware is yet shipping until later this year, but ARM has prepared the 36 patches amounting to 23,000 lines of architecture code for mainline integration."
Well done (Score:5, Insightful)
It's awesome that a major chip manufacturer is willing to invest time to implement a new architecture in the Linux kernel.
Pity that windows isn't open sourced, they wont benefit from this effort ;-)
Re: (Score:2)
Another pity is that (almost) no commercially available devices that will implement this chip will actually run a free OS....
Re: (Score:2, Insightful)
This may help though. This means that much less time and investment is needed to get your new device up and running, providing a convincing a convincing case for switching to free software. Had Linux been ported later, this advantage would have been lost.
Re: (Score:3)
He means that all the ARM hardware will be shipped with Windows 8 and SecureBoot enabled and locked to only run Windows 8 (actually, this isn't strictly true - Microsoft only dictates that SecureBoot be on and locked enabled on ARM, but if you could convince an ARM manufacturer to ship with a Linux SecureBoot signature, then you could still run Linux on it).
Re: (Score:2)
Re: (Score:1)
FYI: Stop buying some cheap Chinese shit. Buy Google reference devices, so you can upgrade them later.
And btw Android is under Apache license, for the most part.
Re: (Score:2)
Google reference devices ARE cheap Chinese shit. Just because they make the die-cast case in the USA and also screw the device together here, does not mean it is made in the USA. Most of the thing is made by Foxconn.
Re: (Score:2)
Re: (Score:2)
in my own family we've gone through 3 Android phones so far that couldn't be updated and we aren't even a smartphone heavy family, if we went through 3 in 2 years i can imagine how many more are sitting in sock drawers right now because they can't run the new version.
That's why for the premium Android experience I always suggest people buy the Nexus device. I have a Nexus S and a Galaxy Nexus and both have the latest Android and have been guaranteed to get the jelly bean treatment. As far as I know the Galaxy Nexus is available for all major US carriers as well.
Re: (Score:2)
That's why for the premium Android experience I always suggest people buy the Nexus device. I have a Nexus S and a Galaxy Nexus and both have the latest Android and have been guaranteed to get the jelly bean treatment. As far as I know the Galaxy Nexus is available for all major US carriers as well.
The verizion Galaxy S is locked down
http://www.xda-developers.com/android/no-root-for-verizon-galaxy-s-iii-petition-started/ [xda-developers.com]
Re: (Score:1)
The Samsung Galaxy S and the Samsung Nexus S are different devices.
The Nexus S is a Google branded device and it is not locked down.
Re: (Score:3)
Re: (Score:3)
ah yeah they will just not for consumers it will be used in linux servers
Re: (Score:2)
HP could go a couple ways yet. They've got a 'softy leading them right now.
But this is interesting tech. [hp.com]
One HP Guy Once Told Me Something (Score:1)
HP does not use AMD processors (at least in serious numbers), because that would threaten their cozy relationship with Intel. If they were "unfaithful", they would not get engineering samples of new CPUs early. That HP guy was even happy about that and pointed out that "IBM is always late to have new Intel processors as they use AMD CPUs".
That is the HP mindset - bend over to get a good assfuck and even be Very Happy to have Intel's dick in their digestive system. Will they ever seriously push ARM-based ser
Re: (Score:2)
Re: (Score:2)
I bet Gumstix will support 64-Bit ARM (Score:2, Informative)
I have a Gumstix Overo Fire COM - "Computer On Module". It really is about a size of a stick of chewing gum, however the I/O board it mounts on is much bigger. I'm heavily into woodworking, so I'm planning to make a real nice hardwood case for my Gumstix Android Tablet.
Gumstix sells individual units to hobbyists, but most of us have commercial products in mind, at which point Gumstix offers volume discounts.
The schematics of the I/O boards are Open Source.
Mich [dulcineatech.com]
Re: (Score:2)
Great links. Thanks for tipping a nerd into the right direction. Unfortunately you make exactly my point. To go from a gumstix module to a tablet or wearable device is a long, long way. Long enough for me too to start pondering if it is worth it to dump my next year's free time onto creating one single device (not even pondering the costs).
Re: (Score:1)
The Kindle 1 team started development using gumstix boards. Once they got the software to a certain point, they were able to make better decisions on what hardware to use in the released product and had a custom board built. (not based on gumstix at all)
Re:Well done (Score:5, Informative)
Don't be so sure - first to market is a major factor in business and if Linux is likely to beat all other rival OS' by a large enough margin in time, commercial vendors will look at that very seriously. More than a few would likely "gamble" (*cough*) on a free OS and gain marketshare when the profits are high than risk coming in very late when there's much less money floating around, a much higher entry fee and customers unhappy with them being late to the party.
It is, of course, essential that the chip works (remember Transmeta?), but hardware sells when there's software and if there's Linux support then there's software - and a lot of it. Assuming nobody has messed up, the chip is going to get deployed. The question is only one of where. Phones, yes, but not necessarily immediately as a lot of apps are compiled natively (not to an intermediate form) and the market is crowded with patent trolls right now.
Re: (Score:3)
Well, sort of.
All mobile devices I have seen with a modifiable bootloader weren't what you would call top of the line products.
And there exactly lies my rant in. All mobile devices worth working on (except the now abandoned n900 and n9)
are locked down into some sort of proprietary eco system, yes that's right I'm calling Google a proprietary eco
system.
It's plainly disappointing.
Re: (Score:2)
No. That is completely wrong.
No android device to date has had a completely end user programmable boot loader.
Re: (Score:2)
Re: (Score:2)
I'm pretty sure that within an hour of it being public, somebody was working on it.
But not me.
Re: (Score:2)
No.
There! Cited. ;-)
Re: (Score:2)
Re: (Score:2)
It is funny to be seeing dozens of ACs respond to my comments today but almost no actual humans.
And it isn't even as if the protrayed responses are offensive to the comunity, excep for being misleading of course.
Re: (Score:2)
Re: (Score:2)
I just can't get over the fact that even though Android "flourishes" as a FOSS project nothing is actually brought back to the rest of the FOSS community. More so FOSS has started to be drained of resources by more and more people jumping on the Android bandwagon. In stead of adding to the wider community Android is fastly moving to make it a mono culture.
At least meego, maemo and mer were giving back to or improving their respecitve projects.
Re: (Score:1)
Re: (Score:2)
IIRC the kernel alone won't help you much with that. You would still have to run a big part of android itself. Also, I think most applications that are not device dependent (not location aware apps, not radios etc.) always had better counterparts in Desktop Linux land anyway. The last bit obviously is debatable.
Re: (Score:2)
Re: (Score:2)
Forget the article.
You are 6 levels abstracted from that.
Re: (Score:3)
Not true. nvidia's project Denver [wikipedia.org] will have to run a free OS if it is really going to be available for "personal computers, servers, and supercomputers"
Re: (Score:2)
Another pity is that (almost) no commercially available devices that will implement this chip will actually run a free OS....
Huh? What are you smoking? This thing will probably run nothing but Linux.
Re: (Score:2)
I'm certain there will be lots - 64-bit isn't currently too important for the consumer space (really, you need 4GB of RAM in your phone?).
No, the real benefit will come from the high end stuff - servers mostly. And you can bet that servers run Linux. Now there's a lightweight lower-power machine that can probably do 90% of the server tasks Sure it won't run a DBMS in most cases, but
Re: (Score:2)
Well, tbh when I first composed that comment I was actually only thinking of mobile devices.
Truth is arm servers could do great as caches or file servers.
Re: (Score:3)
Of course MS will benefit. Their ARM Surface computers will fail due to lack of win32 x86 compatibility.
2015 will be the year of Office on 64bit Android.
Re:Well done (Score:4, Insightful)
Do you really think Microsoft provides their kernel source to ARM? Do you REALLY think ARM would code hardware support in windows for free?
Yes, and yes. It's in both their interests to do so, Microsoft's, because it gets them kernel support for a new arch written for free, ARM's because it sells chips (and by extension, chip designs).
Re: (Score:2)
Re: (Score:1)
You cannot be further from the thruth.
First time I've seen Windows ported to ARM was almost 8 years ago, with a short demonstration by a Microsoft engineer.
They had done all the porting effort themselves (IIRC it was written Longhorn build x on bottom of the screen).
Only thing I learned during this session was issues due to lack of LDREX64 support in ARM ISA for semaphores (which was added right after), and issues he faced with page tables synchronisation (ie. Data cache vs MMU).
Been here, seen that.
Re:Well done (Score:5, Informative)
Re: (Score:2)
That makes it even more awesome. AMD actually makes their own shit and yet they never bothered to contribute proper power saving for (among others) Athlon 64 L110 for Linux. They haven't given out proper support for R690M chipset or the graphics in it which are referred to as X1250. That ARM holdings is willing to do what AMD apparently can not or will not do makes me appreciate them all the more simply for recognizing what can and should be done.
Re: (Score:3)
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
If they don't manufacture physical objects, and the companies are "only buying the specs" then how do ARM processors get built?
Re: (Score:3)
There are two kinds of ARM licensees. One is purely a block licensee - these guys basically buy the ARM RTL ready to be synthesized and plopped on silicon (or tested on FPGAs). Most ARM licensees go this route.
The other is a microarchitecture license. There are only 3 known ones - Marvell (acquired from Intel who got it from Compaq, who got it from DEC), Qualcomm, and Apple (who act
Re: (Score:2)
Re: (Score:3)
Pity that windows isn't open sourced, they wont benefit from this effort ;-)
I assume porting to the NT kernel would require virtually re-writing them from the ground up to fit NT's structure, so not much lost there.
Re: (Score:2)
At one point the NT kernel was on i386, Alpha, and PowerPC. Later ports had AMD64 and Itanium support. With Windows CE/Windows Phone, and the Zune OS, MS has a lot of experience with multiple chipsets. I don't think they would have as much of a problem as you think.
Re: (Score:2)
Re: (Score:2)
They basically realized that the possible revenue warrants the investment.
Good business sense, a rare sight these days.
Awesome... 23k lines gives support to a new arch (Score:4, Interesting)
I think that what is really awesome is that adding just 23k lines of code gives you support for a new CPU architecture!
Bandwidth? (Score:2)
But will they encourage the implementers (chip makers) to create chips with architecture to truly support 64-bit computing. At the moment, most ARM architectures might as well be 16-bit when it comes to bus bandwidth and data transfer. No use having high speed RAM of 512-2048 MB when the interface to the storage (typically flash) can't even touch SATA 1.0 speeds.
Seriously, ARM would do more good for itself pushing vendors to adopt proper multi-channel PCIe (>x4) in their architectures to multiple devic
Re:Bandwidth? (Score:5, Informative)
And even if it is not common in today's products, there are a lot of recent high-level ARM SoCs that offer SATA - not least because its low pin count makes it easier to route on the board in the end than a parallel bus. For example, TI's OMAP5, Freescale i.MX53 or CSR's Prima 2 have SATA support.
Re: (Score:3)
well, there's plenty of use to having lots of memory, especially because the storage is slow.
if you'd like to start using them for clustered db's etc, then the memory is very useful.
afaik it's 48bit memory addressing though what it supports.
Re: (Score:3, Informative)
afaik it's 48bit memory addressing though what it supports
For the record, it's also what most current AMD64 implementations support: https://en.wikipedia.org/wiki/X86-64#Architectural_features [wikipedia.org]
Re: (Score:3)
Freescale is working on such a chip: http://www.eetimes.com/electronics-news/4375606/Freescale-adopts-ARM-cores-in-QorIQ-line [eetimes.com]
Re: (Score:2)
Even for 32-bit instructions, the bus to physical memory can be 32, 64, or 128 bits wide. The reason is that on-chip clocks to cache can be 10 times or more higher than external memory. PCI is not the only or fastest bus system.
Also the ARM-based server chips do have PCI and SATA.
Get ready for the next wave of marketing... (Score:2)
In 2013, the new cool thing to have will be a 64-bit processor! Like in the good old times of the console wars.
Seriously though, in the near future the amount of available address space to be shared between userspace, kernel, GPU etc. might start to become too tight in 32 bits even for smartphones, at least the biggest ones.
Great! Now, can we all switch to 64bit, please? (Score:2)
The company that I work for (non-IT) just decided to upgrade from WinXP to Win7, but they are still sticking with 32bit! What an insane decision. This means the lease of more than 10.000 brand new computers that will stubbornly cling to the past by refusing to make the step to 64bit. I had to raise my voice significantly, explicitly stating that I will not be able to do my work unless I get a 64bit machine with a 64bit OS (which is true). I finally got it, but I guess the folk down at the IT department all
Re: (Score:1)
It turns out that a 32 bit windows patched to enable PAE runs better than a 64 bit windows on the same hardware. I have tested it both ways.
Re: (Score:1)
I finally got it, but I guess the folk down at the IT department all know my name and hate me for not sticking with the rules.
As the should dislike all trouble makers.
Re: (Score:3)
Aside from the memory thing, which you dont need to run IE, and Office, I have yet to see a significant reason to run 64 bit windows clients, and it usually saves a little bit of money as well. So unless you want to pay for it (and how did you do your work on a 32 bit win xp machine but now need a 64 bit os?)
Now linux 64 bit is about as useful as windows XP 64, you want a throwback, hardly anything works out of the box and you spend your time hand compiling what seems like every piddleshit thing. Now grante
Re: (Score:2)
That's the complete opposite of my experience with 64-bit Linux -- everything has worked as well (or not) as it did on 32-bit Linux. Why shouldn't it? Obviously the code needs to be recompiled, but generally someone else does that, or else I'd need to do it myself on 32-bit anyway.
On Windows 64-bit we've had many problems with odd (often old-ish) drivers not being supported, and old software being buggy, even in "32 bit mode". So far, we only have about 15 machines running 64-bit Windows 7, but all the L
Re: (Score:2)
Lots of stuff works in x86_64 in Linux. What problems are you having?
Re: (Score:2)
how did you do your work on a 32 bit win xp machine but now need a 64 bit os?
I didn't. I used a 32bit XP box for Office-work and internet browsing and I logged in to a 64bit SUSE sever via PuTTY for doing serious work. But then I bought a 6-core desktop computer with 24 GB RAM and, of course, I just couldn't let the IT people just slap their 32bit Win7 image on it. It would just beat the purpose of having such a machine.