First Debian/Ubuntu Bootable ARM64 Images Released

An anonymous reader writes "With work done by ARM and Linaro, there is now a bootable image of Debian/Ubuntu that works for ARM64, the new 64-bit ARM architecture. There are still some caveats and work ahead, but Linux is once again the first platform that has software ready to run on a new architecture when released. This 64-bit ARM Linux support also includes the ability to run 32-bit ARM software side-by-side." You can grab a bootable rootfs, but there's no hardware to actually run it on now (the developers are using the free-as-in-beer simulator from ARM). Kernel support for the architecture was released around a year ago; this is more a tale of getting from a bootable kernel to a bootable operating system.

Re:

[Anonymous Coward]

It might happen, but if it did it wouldn't be very smart nor particularly interesting. ARMv8 is firmly targeted at the data center: hence the need for 64bit support and the interest in making sure Linux supports it out of the gate.

Software/hardware

[gTsiros]

Must be one of the few times in history where the software was ready to run before the actual hardware existed.

Re:Software/hardware

[fuzzyfuzzyfungus]

I think that the same was true when AMD was first pushing x86-64 extensions.

And probably more common than that: If you are going to a chip to implement some particular function, you need to have a very good idea of what you want already in mind. And, if you have one of those, there really isn't any reason to not have the compiler guys begin their work, or any significant obstacle that would keep somebody from hacking together an emulator implementing the ISA before the hardware team manages to get production silicon in play.

Writing emulators that are 100% accurate in simulating every last quirk of real hardware, and ideally doing so at useful speed, isn't easy, nor is producing compilers that reliably produce good output that takes advantage of the real strengths and weaknesses of real chips; but producing merely functional versions of both based on the spec faster than the hardware team can produce a good silicon implementation of the spec is likely a winnable race most of the time(especially now that 'just throw a dirt-cheap and sickeningly powerful x86 at it' is a viable strategy for papering over issues with your emulator).

Re:

[gTsiros]

Yeah, but this time, the ISA is complete, the *actual hardware* though doesn't exist and it is not very likely it will change a lot (besides bugs and optimisations, i would guess).

so the design exists, the software is ready for it (as ready as it can be, considering that only the specification exists) but a hardware instance doesn't yet exist.

Sorry i can't describe it any more accurate, english is not my mother tongue.

Re:

[K. S. Kyosuke]

Yeah, but this time, the ISA is complete, the *actual hardware* though doesn't exist

What the hell are you talking about? Live demo of 64-bit Linux on ARMv8 *in silicon* was presented a year and a half ago!

Re:

[superzerg]

I believe the first games for any consoles have been developed on emulated machines

Re:

[Sarten-X]

Any consoles? Including Pong?

Re:

[RoboJ1M]

Pong wasn't written, that was back in the day everything was practically soldered
So it probably really was done on a breadboard.

Re:

[interval1066]

Spacewar [wikipedia.org] is probably a better example.

Re:

[T-ice]

they also eat sausage...

Re:Software/hardware

[oodaloop]

Following the fine tradition of writing software going all the way back to Ada Lovelace and Charles Babbage. She wrote the rules how the machine should work long before the gears were even made.

Re:

[Lennie]

And so Dijkstra still did it.

Re:

[K. S. Kyosuke]

Must be one of the few times in history where the software was ready to run before the actual hardware existed.

You make it sound as if CPU manufacturers aren't in the habir of validating their architectures before fabricating them, by running actual programs on a simulated device.

Re:

[Kjella]

And even more so on the GPU side, CPUs may add a few instructions each round but GPUs rewrite the book quite often, here's from nVidia's hardware emulators [epmag.com].

Re:

[Anonymous Coward]

The very first ARM was developed and simulated on a 6502 based BBC micro by the Acorn team.

Re:

[Kidbro]

Nah. None of the software I write runs on existing hardware. I prefer to think of it as being ahead of the curve!

That's amazing!

[ArturoBandini77]

The first OS that can run on hardware that's not released yet...
Someone remembers USB support in Windows NT???

What's Debian/Ubuntu?

[Compaqt]

Is that like GNU/Linux? Debian saying "we should have our name in there, too"?

Re:

[Ginger Unicorn]

Nothing in his post implies that he doesn't already know that.

linux is first?

[YesIAmAScript]

'Linux is once again the first platform that has software ready to run on a new architecture when released'

It's the first you've seen. That doesn't mean it's necessarily actually first. It's just linux "shows its work" as it goes along. MS is not going to do the same.

It's great to see multiarch rolling along.

There is hardware

[Anonymous Coward]

There is hardware:
http://www.theregister.co.uk/2013/01/17/applied_mixcro_x_gene_open_compute/ [theregister.co.uk]

We can't buy it yet but it's in the lab.

Not all that surprising

[MOSFET Explosion]

Since ARM64 is going to be targeted for tablets an smartphones, and since a large portion of the tablet and smartphone market right now is Android. It's really not that surprising that the Linux kernel would be ported over so quickly.

That's good...

[UltraZelda64]

...but how long until Microsoft swoops down, releases a Windows version for the platform, and requires Secure Boot to be enabled with no way to turn it off--effectively locking Debian and Linux in general back out?

And yet it won't get hardware support.

[Mabhatter]

So the chances that we'll get an upgraded barebones like Raspberry Pi with the new processor runnin BARE Linux, and not with super-secret DRM?