Slashdot Mirror
ARM Launches Cortex-A5 Processor, To Take On Atom
bigwophh writes "ARM launched its new Cortex-A5 processor (codenamed Sparrow) this week, and while it's not targeted at the top end of the mobile market, it is a significant launch nonetheless. The Cortex-A5, which will likely battle future iterations of Intel's Atom for market share, is an important step forward for ARM for several reasons. First, it's significantly more efficient to build than the company's older ARM1176JZ(F)-S, while simultaneously outperforming the ARM926EJ-S. The Cortex-A5, however, is more than just a faster ARM processor. Architecturally, it's identical to the more advanced Cortex-A9, and it supports the same features as that part as well. This flexibility is designed to give product developers and manufacturers access to a fully backwards-compatible processor with better thermal and performance characteristics than the previous generation."
The Cortex-A5 is aimed at phones. The Cortext-A9 is the one aimed at netbooks. The article referenced in the summary makes this clear.
I would love to have one of these in a "smartbook". Even though it won't run x86 binaries (I use linux anyway) it would be useful enough to let me leave my big arse laptop at home. With hours of battery life I wouldn't need to take a power supply with me.
So far though the only ARM smartbooks currently available have very limited RAM and disk space. I will have to wait and see what comes out in the next few months.
Unicode in Slashdot
The Cortex-A5 is a slight improvement over the MPCore/Arm11/Arm9. That's nice for those who need it, but it's miles away from the speed of a Cortex-A9, which is really what's going to be needed to battle Atom.
And since the A9 has announced by ARM quite some time ago, this posting should have been written then not now.
In reality, it's not clear which niche the A5 is going to occupy. It's probably going to be useful in lower end smartphones only, since current higher end models are already using the faster A8.
Looks like the Cortex-A5 has 50% more performance while using 1/3rd the power of the current generation ARM11 found in the iPhone. As a game developer this makes me hopeful that we'll see cellphones as a gaming platform without sacrificing useful battery life.
---k--
</stupid>
So this is why ARM and Global Foundries recently made a deal. ARM's Cortex-A5 is going to be built on a 40nm and Global Foundries already has that equipment, with AMD working hard to advance to the next node that frees up a lot of manufacturing power for ARM to use. Officially it was for Cortex-A9 at 28nm but what's to stop other stuff from being done in the shadow of the deal?
Its the Wifi/WWAN chips, and LCD screen which suck up the power, not the CPU. ARM is cool and all (pun intended) but if you make an ARM based Dell Mini 9, you're not going to end up with uber battery life, when you're on Wifi and running the screen bright.
I want to delete my account but Slashdot doesn't allow it.
It's said that Intel has the edge on this fight due to x86 compatibility, but Microsoft can really change things around if they decided to port Win7 to ARM, instead of offering only Windows CE. But considering monopolies, I wouldn't expect that any time soon.
ARM talked about the Cortex A9 (the one I'd actually like to have in a netbook) over two years ago. There is still nothing you can get that actually has one in it. Yay something to replace the ARM11. Hope it actually gets used.
The Cortex-A5 has a more advanced L2 memory system with multiple outstanding transactions. This makes a huge difference for many workloads compared to the ARM11 cores. Thus, for workloads not contained entirely within the L1 memories the Cortex A5 should offer much better performance.
So, Sparrow, we meet again.
Yes. Sometimes I think that I am getting too old for this game.
-- The Crepes of Wrath --
Architecturally, it's identical to the more advanced Cortex-A9
How can it be identical, when it's more advanced? Those two are opposites.
Or is their definition of identity itself more advanced? ^^
Like "(==) a b = a >= b" in Haskell?
Any sufficiently advanced intelligence is indistinguishable from stupidity.
Now, ARM -- another British invention -- has established a small beachhead in the notebook market (which includes netbooks). Can ARM do what SPARC, MIPS, Precision Architecture, and PowerPC failed to do? Can ARM actually reach 50% of the processor market for notebooks -- and eventually desktops?
I hope so. I admit that I am biased and love cheering the underdog.
Intel developed its x86 architecture by pumping globs of monopolistic profits into research and development. Too, the massive federal funding (via university research grants) and corporate funding furthered the development of both SPARC and MIPS. By contrast, ARM was developed on a shoestring budget. The goal was modest: low power and average performance.
All the American processors are Goliaths. ARM is David. I hope that David slays the biggest Goliath: x86.
Cheerio.
We really have to start looking more carefully at posts like this, which clearly contain entire paragraphs of unexamined assertions by company PR drones that may or may not be true. Bottom line: Kill this shit unless a trustworthy, honest reviewer with a decent track record says it. If that isn't happening, quit posting it here, where we have more important stuff to spend time on.
By the way, that "more important stuff" includes pulling our dicks and/or replaying World Championship Monopoly games move by move.
I've calculated my velocity with such exquisite precision that I have no idea where I am.
http://arstechnica.com/gadgets/news/2009/10/arm-fills-out-cpu-lineup-with-cortex-a5.ars
http://www.brightsideofnews.com/news/2009/10/21/arm-announces-cortex-a5-for-the-next-15-billion-cellphones-and-mids.aspx
Late? They said 2010 in the article you linked.
In this article, they said Cortex A5 in 2011.
As a developer for products based on ARM9 and ARM11 SoCs the A5 is targeted squarely at me. I'm not sure why it's of any interest to slashdot. But it does appear to be a cheaper ARM11 (to the point of making the ARM9 obsolete) but with some of the features of the A8.
While smartphones are all sexy and exciting, the staple for cell phone manufacturers are the simple ordinary phones. If they can cram more features into the same cheap phone it usually means they can sell more of them. Think of it as competing in the free phone market. Where the styling and brand and features are the only way to differentiate yourself rather than price. The customer is just going to pick 1-4 of the plan bundled phones.
“Common sense is not so common.” — Voltaire
You have to expect pedantry, this is Slashdot.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
Define 'you'. ARM began selling Cortex A9 licenses a while ago, but ARM does not produce chips. TI are shipping OMAP4 SoCs based on the A9 to high-volume OEMs for a little while, as have a couple of other ARM licensees. They should be appearing in consumer products in 2010. As, in fact, it said in the article you linked to.
I am TheRaven on Soylent News
Before the A series, ARM haven't really designed any new processors since Acorn Computers died in 2000/2001. The only development push ARM had is when RISCOS went to other manufacturers such as Castle. Now ARM needs to design new processors as their time has come where more powerful CPUs are needed in the mobile devices.
And it's full of misinformation:
1) The A5 is not meant to take on Atom. The A9 is.
2) The A5 is not architecturally identical to the A9. The A9 is an in-order, multi-issue core. The A5 is an out-of-order, single-issue core. The only thing similar is it has the Cortex A-series ISA.
What the A5 is is a CPU that completely obliterates the ARM11-derived cores, used in everything from NVIDIA Tegra to the Nintendo DS. It's an update of the ISA, and a more capable core, with better thermals. That's it. Whereas every low-end smartphone now has the same damn QualComm ARM11-based core, in a year, they'll all have the A5.
Lex orandi, lex credendi.
ARM11 launched in 2002. That's a pretty major one...
(And, Acorn as a personal computer manufacturer died in 1998. They were using the DEC StrongARM, which predates the ARM9 and ARM10 - the StrongARM was used in place of the ARM8 that was still under development, and the ARM9 borrowed ideas from the StrongARM.)
Actually if Qualcomm has their way all the smartphones will be running a Qualcomm Snapdragon with a Qualcomm Scorpion CPU, their superpipelined version of the Cortex A9.
A Snapdragon should run at 1 GHz (Cortext A9 is 600 MHz on a comparable process), from what I've read the A5 will be 480 MHz on a 40nm process.
So the A5 is aimed at cheaper devices than the Snapdragon. Of course the A5/A9 are presumably available to all ARM licensees whereas the Snapdragon is as far as I can tell only going to be manufactured by Qualcomm.
echo -e 'global _start\n _start:\n mov eax, 2\n int 80h\n jmp _start' > a.asm; nasm a.asm -f elf; ld a.o -o a;
Oops. The Snapdragon is a superpipeline Cortex A8. It's not really clear how an Cortex A9 would compare against a Snapdragon.
echo -e 'global _start\n _start:\n mov eax, 2\n int 80h\n jmp _start' > a.asm; nasm a.asm -f elf; ld a.o -o a;
The A5 isn't competing with the Atoms. It's meant to replace the ARM9s and ARM11s found in a lot of devices from phones to the Nintendo DS.
Mada mada dane.
Plus, it could even take advantage of the enormous number of open source programs that could be compiled for ARM Windows before commercial titles get ported.
Most open source desktop apps that I've seen either are ported to GNU/Linux (e.g. Firefox and OpenOffice.org) or came from the GNU side of the fence in the first place (e.g. GIMP and Inkscape). So Windows NT for ARM wouldn't have a huge advantage over Ubuntu in this case. It would probably be more productive to consider a compatibility layer from Windows CE to Windows NT, much like the Win16 to Win32 and Win32 to Win64 layers that Microsoft has already implemented in Windows NT, so that at least a user's collection of Pocket PC apps will still work.
Making a C program 64-bit safe, if it was not designed to be portable originally, is a lot of effort. Porting a C (or C-family) program from x86 to ARM is generally a straight recompile.
Plus the price of a hostile takeover of the non-free program's copyright owner, which otherwise declines to do this recompile in the interest of maintaining the market segmentation between the smartphone editions (Windows Mobile, iPhone, etc.) and the desktop edition of a program.
But, really, a port of Autocad is irrelevant.
AutoCAD was used as an example. There are plenty of other non-free programs for Windows that won't be recompiled on ARM.
Converting 32-bit code from one CPU to another is much easier than going from 32-bit to 64-bit, so it wouldn't take very long for vendors to update their software for it.
Unless the vendor declines to do the port at all for business reasons. This happened back in the days of NT 3, which was ported to MIPS and PowerPC but most apps still had to run in the emulator.
Also, Microsoft strongarms ISVs into compatibility.
Don't you mean Microsoft XScales ISVs into compatibility?
For example, it's often hard (or harder) to get "Windows Logo" certifications for software unless it works on various platforms.
Does Microsoft demand that all PC games in the "Games for Windows" brand get ported to Xbox 360? No. Desktop PCs and mobile phones are at least as different as desktop PCs and video game consoles.
You can bet that an ARM version of Windows would be accompanied by an ARM version of Office
But how easily would Microsoft Office (for Windows 8 ARM Edition) run third-party extensions designed for Microsoft Office (for Windows x86)?
Most software, unless it uses inline assembly or SSE / MMX intrinsics, is a straight recompile.
A lot of programs' file formats depend on details of the x86 ABI because the programs pretty much just fwrite() a struct to disc.
If you look at the ARM press wibble, then you'll see that they make a distinction between 'smartphones,' which are things like the iPhone, Palm Pre, N900, and so on, and 'feature phones' which are what everyone else thinks of as smartphone (smaller screens, but capable of running user-installed apps, come with a web browser, may support WiFi + SIP, and so on). The A5 is aimed at the feature phone and 'dumb phone' markets, the A8 and A9 are aimed higher.
The big advantage of the A5 over the ARM9 and ARM11 cores used in these phones currently is that it supports exactly the same instruction set as the A8 and A9. That means that you can run the same OS on your entire phone lineup and same basic suite of userspace programs, just adding extra ones on the ones with the faster CPUs. The big advantage that the A5 has over the A8 and A9 (including the Snapdragon) is that it is under half the price. A $40 A8 SoC is not much in something like an iPhone, but at the (much larger) lower end of the market that is getting on for half of the sale price of the phone.
Oh, and the A8 goes up to 1GHz. Things like Freescale's i.MX515 ship at this speed. The A9 is designed to scale up to 2GHz. The Snapdraggon is quite a nice A8 implementation, but it's not particularly exceptional.
I am TheRaven on Soylent News
Oh, and the A8 goes up to 1GHz. Things like Freescale's i.MX515 ship at this speed. The A9 is designed to scale up to 2GHz. The Snapdraggon is quite a nice A8 implementation, but it's not particularly exceptional.
The Snapdragon is not an A8 implementation - it's a custom implementation of the ARMv7 instruction set. Implemented on the same process as n A8 it will run at a 50% faster clock speed and running at the same clock speed it will consume 50% less power.
http://www.insidedsp.com/Articles/tabid/64/articleType/ArticleView/articleId/238/Qualcomm-Reveals-Details-on-Scorpion-Core.aspx
Back in 2005, Qualcomm announced that it had licensed the ARMv7 instruction set architecture and was working with ARM to create its own high-performance core based on that architecture. The new core was dubbed "Scorpion," and at the time it was announced, Qualcomm didn't disclose much about it except that it would run at 1 GHz in a 65 nm process and would be customized to provide a high level of performance and energy efficiency in its target mobile applications. Exactly how this combination would be achieved was not revealed, which is typical of Qualcomm; historically, the company has disclosed few details about the processor cores that live inside its chips.
Then in 2006, Qualcomm announced a new chip platform, "Snapdragon," in which the Scorpion core would be used alongside several other processors and co-processors. According to Qualcomm, Snapdragon will serve a range of high-performance mobile applications, such as high-end smartphones and mobile internet devices. Still, there was little information about the Scorpion core itself.
In conference presentations this year, however, Qualcomm popped the hood on the Scorpion core and presented a detailed description of the core's microarchitecture and implementation. The Scorpion core is similar to ARM’s Cortex-A8, which also implements the ARMv7 architecture. Like the Cortex-A8, Scorpion is a superscalar, dual-issue machine, and supports the powerful, signal-processing-oriented NEON instruction set extensions and VFPv3 floating-point extensions (referred to collectively on Scorpion as the "VeNum" media processing engine). Scorpion will be supported by ARM's standard software development tools, and Qualcomm expects to offer off-the-shelf multimedia codec software that uses VeNum.
Although Scorpion and Cortex-A8 have many similarities, based on the information released by Qualcomm, the two cores differ in a number of interesting ways. For example, while the Scorpion and Cortex-A8 NEON implementations execute the same SIMD-style instructions, Scorpion's implementation can process128 bits of data in parallel, compared to 64 bits on Cortex-A8. Half of Scorpion's SIMD data path can be shut down to conserve power. Scorpion's pipeline is deeper: It has a 13-stage load/store pipeline and two integer pipelines—one of which is 10 stages and can perform simple arithmetic operations (such as adds and subtracts) while the other is 12 stages and can perform both simple and more complex arithmetic, like MACs. Scorpion also has a 23-stage floating-point/SIMD pipeline, and unlike on Cortex-A8, VFPv3 operations are pipelined. Scorpion uses a number of other microarchitectural tweaks that are intended to either boost speed or reduce power consumption. (Scorpion's architects previously designed low-power, high-performance processors for IBM.) The core supports multiple clock and voltage domains to enable additional power savings.
In addition to developing a custom microarchitecture, Qualcomm also customized the core's circuit design and layout in an effort to improve energy efficiency.
Overall, Qualcomm has made a huge investment in creating a custom implementation of the ARMv7 architecture. By way of comparison, Texas Instruments customized just the
echo -e 'global _start\n _start:\n mov eax, 2\n int 80h\n jmp _start' > a.asm; nasm a.asm -f elf; ld a.o -o a;
Has anyone found intelligently done benchmarks which pit Cortex A9-MP against Intel Atom?
Quite literally, I think you'll find.
and the only advantage of Netbook over a real mini communication device which has OS designed to run on it is? Ability to run Windows unmodified.
I speak about devices like Nokia E90, N97, N900.
ARM really gets confused very easily it seems. They should ask Asus, HP, Acer and several others. Why does a customer buy a low powered laptop like device for?
Speaking (typing) from a Quad G5, PPC and watched the happenings in OS X community/developer scene since Intel transition announced. If Intel one day manages to make Atom (x86) run in same low power as ARM licensed CPUs, ARM is doomed.
Why? Compare the compile process of an open source, multimedia application on PPC and Intel. See the "bonus" stuff Intel chips get? Every kind of optimization, way more cheaper is available on Intel x86/SSE. Trust me, I am more amazed to Intel's developer/development/application support every single day. I don't even blame Apple anymore, I blame IBM/Motorola etc.
How about this.
Let's define a constant as X.
There is X ARM software available
There is at least X^Y where Y is >= 2 x86 software available
Seeing as I have no idea what the actual numbers for ARM or x86 software is I decided to express the relationship between the two functions for any definition of X and Y as is accurate to the data.
In this case you ought to say "there is an order of magnitude more software available...(base Y)".
Otherwise, using your argument, for the same numbers you could equally say "there is a constant factor more more software available", in that for X ARM apps there are X*c x386 apps for c > 1
Obviously this is true for the constant c = X^(Y-1)
As soon as you start playing with two constants, there is an infinite number of functions that can interpolate between them (and hence describe their relationship). This includes exponential functions, polynomial, constant, sinusoidal, linear...
But ARM has those spiffy DSPs. More and more codecs are going GPU or DSP powered, so who cares about CPU optimizations for such multimedia tasks?
By the time an Atom has as low power consumption as an Arm processor, Arm processors will be faster. :/
The DS does not use ARM11, it uses one ARM9 and one ARM7.
I'm shocked at this claim. Back in the day, Byte Magazine used to dissect processor architectures in a way you rarely see any more, apart from anything written by Jon Stokes over at Ars. Realworldtech picked up the torch, and I followed it for a while; smart guys, but you need a large Kool-Aid division factor to hang there.
This problem of "true innovation" has dogged the computer industry since the introduction of Hype 1.0.
Kurweil's law is "no technology before its time". Why is it that the premature ejaculator so often gets the lion's share of the credit? You can't deny the innovation at Xerox. The Xerox Dorado from 1979, which I once used for an hour, is reputed to have contained 3000 discrete ECL chips and have a BOM cost pushing up into six figures. Retail price might have been in the $200k range if, say, all the moon rocks recovered by NASA had been made of solid gold, and the engineers were suitably rewarded. I was told my my friend, a coop student there at the time, that the rumour on estimated street price to sell the Dorado was "probably $250k". I thought that was high at the time, but I knew less then about cost multiples.
Ray Kurzweil on how technology will transform us
When you run a giant fab, you need to consider your volume targets in choosing processor design goals. What made the Alpha kick ass was the incorporation of some ultra-expensive metalization. That's how you get fast 64-bit adder in early 1990s process technology: an entire layer devoted to fast carry propagation. Lacking OOO, you need short, deterministic instruction latencies above all else, unobtainium be damned. Works for NASA, Boeing, and Ferrari. This fabrication approach was a total non-starter for Intel volume production.
IIRC--and this is becoming dim--the Alpha was a four-issue core with a uniform instruction width and precious little OOO logic. What is it that Nahalem is reputed to have copied here? It's been known for 15 years now that x86 integer performance was able to directly compete with RISC designs given a large design team devoted to working around the instruction set wonkiness. Most of the problems with x86 were toll bridges, rather than permanent road blocks. On the floating point side, the blighted x86 stack architecture cost you a factor of two. But floating point defined the low-volume workstation market, where sports cars like the Alpha found fleeting glory. I actually think the Itanium better represented Intel's desire to take Alpha to the next level.
Apart from that, over the longer time frame, reality imposes convergent evolution. To my knowledge, Intel never once publicly stated that AMD's on-die memory controller was the wrong path to take. Intel usually said "not yet, we can do it cheaper for another spin without going there, and besides, our marketing department ate some bad mushrooms for a couple of years there, so our roadmap is a bit jumbled right now." Does AMD get credit for innovating on-die memory controllers or for facing up to despe
We need some motherboards with a couple of hyper-transport-3 enabled A9s, and sockets for RAM. At least 8GB (16GB NUMA?). That connected to a semi-decent south bridge (with pci, pci-e, sata, usb et al.)
This board should cost less than USD$150 with the 2 A9s. And after a few months maybe < $100?
That's a dream because the A9s are not working in dual socket configurations yet and ARM doesn't invest in the desktop with a motherboard reference design. Sad...
So you exclude the .NET applications (as .NET would be ported to ARM by Microsoft).
Not all applications written in C++ have been rewritten in managed C++ or another .NET language. Not all developers want their apps to depend on the tens of megabytes in the .NET framework, especially if they still have a lot of users who don't already have the framework installed. Nor does it help apps that use PInvoke to call native libraries that fill holes in the .NET framework.
And you stop considering the system calls (as they would be native ARM anyway).
Can an end user quantify how much time a given application spends in user space vs. the kernel? I have access to a couple PCs running Windows XP; does Task Manager or a similar tool let the user separate out syscalls from user mode? And the last time I looked at a mixed-instruction-set environment (mixed x86/PowerPC in the era of System 7.5), context switches between instruction sets were rawther expensive.
First, 2GHz dual/quad ARM Cortex A9s are required.
I'd like to know how much battery power would be required.
ARM recently released a 2GHz dual-core A9 hard-core for TSMC's process (40nm). I forget the power consumption, but it was quite low compared to Atom.
http://www.electronicsweekly.com/Articles/2009/09/16/46955/arm-produces-hard-cortex-a9-for-high-performance.htm
ah... http://www.arm.com/news/25922.html
"The Cortex-A9 power-optimized hard macro implementation delivers its peak performance of 4000 DMIPS while consuming less than 250mW per CPU when selected from typical silicon."