31 comments

[ 4.5 ms ] story [ 71.7 ms ] thread
Arstechnica has a great write up: http://arstechnica.com/business/news/2010/09/arms-eagle-has-...

Interestingly, Ars says it's a laptop and server chip, not really aimed at ARM's traditional embedded market.

Tablets, likely, laptops very unlikely - neither Windows or OSX run on ARM (which together make up over 98% of laptop OS market share http://gorumors.com/crunchies/operating-system-market-share/). I suppose Windows 8 and OSX 10.7 could be out before 2012 and ported, but I wouldn't bet on it....
who cares about win and macos - it will run on linux.
Is there any good reason to sacrifice any significant amount of battery life for performance in a tablet class device at this point? I think consumers will find the happy medium of device battery life (full day for SmartPhones, couple of days for tablets) and CPU evolution will need to be more incremental to meet those demands.
Honestly, I don't need a computer more powerful than my Motorola Droid. If someone could make one of those that had the battery life of a feature phone, I'd take my Moto Droid any day of the week.

(Actually, I don't see why you think that tablets should have multi-day life expectancies. I'm almost always using them in areas where there are plugs. My phone on the other hand almost never leaves my pocket. )

"Honestly, I don't need a computer more powerful than my Motorola Droid. "

How can more performace ever be a bad thing, things you have not even imagined or been invented yet to do on a mobile because the performance is not there yet. E.g. things that I know: 3D video recording and photography, great performance augmented reality, instant search with social network and location data built it, HDR video recording.

I don't even know what I don't know...

And its not even multiday - I would love my iPhone to be double the performance and 8 hours non stop video play. That would be awesome, nothing more frustrating than being stuck on a train with a mobile that has run out of battery, only so long you can stare at the hot chicks :)

A mobile is a tool. It accomplishes a set of tasks for me. I don't want to be bothered with constantly worrying about my power supply in search of hypothetical benefits. I want a music player / messenger / Internet browser that I can plug into an outlet at my leisure.

Also, instant search with location / social data already works on my Motorola Droid. In fact, certain parts of it work worse on Froyo than they did on the previous release.

(Example: I used to be able to use voice search to say "Navigate to John Smith" and if I had a contact named John Smith with address information, it opened up navigation and took me there. Hopelessly broken on Froyo, and there's nowhere to submit a bug report because voice search and maps are both closed-source.)

But I guess mostly what I'm saying is that there are no hardware problems, it's all cloud data, cloud computing, and the main bottleneck is network, not CPU. Even so, the bottlenecks don't cause enough trouble that I'm willing to sacrifice battery to get rid of them.

the main issue with that for me at least is offline useage. I often use my phone on the London tube with no internet access so having things offline or cached from the cloud is great. Local processing is absolutly required.

LTE and new wireless standards may eventually change this but currently you can't beat something that runs locally. So more powerfull hardware on phones is absolutly required

There's probably things in theory you can do with it, but in practice I really haven't found a need for more performance in the past 5 years or so. I have a low-powered, old laptop and a full-powered, newish desktop, and I really don't run into anything I want to do on the laptop but can't. Admittedly, I don't play much in the way of recent games, so that would be one possible use. There's scientific computing, but I run that all on remote servers or by spinning up VMs in the cloud these days. I just haven't run into much of anything else that needs the CPU cycles. In retrospect, I should've bought a lower-powered desktop machine!
Each to their own I suppose, I have a few laptops generally replace them every 2-3 years. Currently an Alienware M11X and a Acer 1215N. Even for "just" email, browsing, reading I really like the "instant" feel from the newer core 2 duo processors and an SSD.

I also occisionally fire up a bit of Starcraft 2 so the M11x is great for that.

I use an older computer at work and it is really painfull waiting for it to load apps and even Firefox - ultimately it is only a few secords but that is annonying to me.

For my phone the iphone is brilliant but battery life is a real issue. I read on it all the time (walking, trains, tube, waiting in line). Occasional game and youtube video also. Extra battery would be great espeically when I'm on holiday or weekend break or just have played a few videos and haven't had a chance to charge it all day

The performance is fine at the moment but I would love it to be faster - those apps to launch instantly, multitask instant, youtube vids play in HD with no lag and the ability to do more on it e.g. take HDR video, compress and edit movies I take on my camera, play full 3D games with great graphics e.g. upcoming Sword.

You can keep your 5 year old machine, but I am really looking forward to the Dual core ARM processors - thats why I own share in ARM and Qualcomm :)

Hmm, the SSD is a good point. I don't think my interactive desktop usage is very often CPU-bound, which is one reason I don't notice much difference between the older and newer machines I own. But an SSD might make some difference, because I'm guessing some stuff (esp. app startup) is I/O-bound.

I do often take a software approach to speeding things up, though. For example, my solution to Firefox taking a few seconds to start up was to switch to Chrome. ;-) And thanks to the amazing JS compiler advances over the past few years, my old laptop runs JS much faster today than it did when it was new!

Yes buying an SSD was the single biggest increase to speed I have ever seen from one component. I run SSDs on all my laptops and I would highly recommend it. Have a 2TB USB for all my storage so 128 or 64gb SSD which is not that expensive is fine.

Agree on software approach - I love Chromium and really struggle when forced to use Firefox for work. IE I wouldn't touch with a 40 foot pole and thanks the IE tab extension for Chrome I never need to :)

Seconded. If I could get a cheap, smallish (32 GB is enough), durable, and fast PATA SSD for my six-year-old laptop, I'd have everything I need.

As it is, I have nearly everything :)

""Honestly, I don't need a computer more powerful than my Motorola Droid. ""

But the phone does.

Originally the ARM was a tiny, 30,000gate, cpu core in the corner of the ASIC that did all the phone stuff. If the ARM gets powerful enough it can do all the protocol specific stuff and you no longer need an ASIC. So reduced parts count, quicker development, easier to support new phone standards etc.

Android/ChromeOS smartbooks/laptops would be a good bet on the target market for the laptops comment.
For laptops you have to remember that, at this time, the biggest power consumption (for screens bigger that 10-12") is the LCD (mainly his backlight).

So, what you're gaining putting in an high-end ARM against a "low-power" X86 is a portion of the total power consumption of the laptop.

Why choose ARM then?

When the screen real estate is smaller than 10" or the size of the CPU+Logic Chip is important (take a look at the size of an Apple A4 against an Atom+Chipset). And when you don't have a large codebase already available.

The new AppleTV is a good example of this, the usage of the A4 is a no-brainer whatever Intel CEO says: http://www.businessinsider.com/intel-ceo-steve-jobs-took-a-s...

s/laptops/netbooks as atom competitor.

It could allso give you better allways on with a real laptop where you dont need to hibernate to save power.

ARM has a different take on this, positioning this MCU (int the 1 GHz – 1.5 GHz single or dual-core configurations) even for Smartphone and Mobile Computing: http://www.arm.com/products/processors/cortex-a/cortex-a15.p... (under performance tab).

Then you scale up to 1.5GHz – 2.5 GHz quad-core, octo-core or larger configurations but this are obviously for other class of devices.

Intel should be worried. With support for 1TB of memory this seems like the first ARM chip that can really make a dent in the server market. I expect the big players with significant energy bills to be very interested in this.
Supporting 1 TB of memory on a 32-bit system reminds me of the bad old days of DOS extended memory. Are there any plans for a 64-bit ARM chip?
I don't know, but it's the obvious next step. There are certainly rumours that ARM is going to go that way. I'm not plugged in to the chip manufacturer gossip network but my guess is that ARM doesn't want to start an open fight with Intel before they have

1) Tested the server market. A 32-bit chip has its limitations, but for things like caches that are IO limited it is fine, and

2) Got their 64-bit architecture nailed down. I expect they don't want to give Intel a long lead time to respond.

When power is a major concern, and that is the applications ARM is targeting, using a 64-bit system when 32-bit would suffice is counter-productive. A 64-bit CPU naturally requires more transistors and thus more power than a 32-bit one. There are other overheads (in OS etc) associated with a 64-bit system. The difference may not be great, but for many applications it is a cost with no reward.

Comparing to MIPS, they have had a 64-bit ISA variant for a very long time, yet all the recent cores have been 32-bit as this is all their current target markets require.

A full 64-bit system is really required only for HPC-type applications, a domain currently owned by Intel and IBM. If ARM are interested in eating into this market segment, it is most likely a very long-term goal, and obtaining a foothold in the 32-bit server space would serve as a first step in that direction.

Networking equipment is another place where 64-bit systems are useful. See chip offerings from places like NetLogic and Cavium, for instance: their multicore MIPS chips are 64-bit.

Databases are another place where 64-bit systems shine. Maybe you already assumed that was included under "HPC-type applications", though. When I hear HPC, I think number-crunching, not databases, though.

40-bit addressing is probably enough for a lot of applications that Intel isn't a significant player in yet, also -- like console gaming, streaming video, and that sort of thing. That would also explain the target 2.5 GHz clock speed, 8 cores, and media and vector instructions.

I'd also guess that limiting the addressing to 40 bits is partly for power consumption, as mentioned elsewhere.

You're probably right about not wanting to start an open fight with Intel, but it's hard to not see this as a shot across the bow, especially compared to Atom. If nothing else, it should make the competition between Atom and Bobcat more interesting :)

This is nothing like DOS extended memory. It is quite similar to Intel PAE, allowing more than 4GB of physical memory. Each process has the usual 4GB virtual address space; the MMU simply maps it into pages taken from a larger pool. It makes perfect sense when running multiple large processes or doing virtualisation.
I could see ARM processors really threatening the server market. Do you know how much power is used in a typical data center?

In fact Smoothstone just raised a lot of funding just to pursue this: http://www.smoothstone.com/

Also with increased virtualization you have beefier and beefier single servers running 100's of servers / desktops. Imagin running your entire VDI estate on a single 16 core blade powered by ARM processors that uses as much power as a single desktop, can you smell the $$$

Where does one get a datasheet for one of those?

I was looking for the datasheet for previous models, but all I could find were press releases.