CHINESE CHIPMAKER Nufront has managed to implement ARM's dual core 2GHz Cortex A9 chip in one of its system-on-chip (SoC) designs.
The 2816 Computer System Chip is, like all of ARM's designs, extremely power efficient with Nufront claiming that the dual core Cortex A9 design sips just 2W when running at 1.6GHz. The firm is heralding the chip as an opportunity for it to go up against Intel in the ultra-thin laptop and netbook market.
Nufront claims to be the first company to integrate ARM's dual core Cortex A9 chip at 2GHz. Other parts of the SoC include a multi-core graphics chip, a 64-bit DDR2/3 memory controller, a 1080p "multi-format video engine", SATA2, USB2, Ethernet and a general I/O controller.
ARM's architecture does not natively support the x86 instruction set, meaning that a bog standard copy of Microsoft Windows will not run on ARM chips. Nufront said that it is working on getting the open source operating systems Ubuntu and Android to run on its SoC.
Nufront is the first Chinese company to use the 40nm manufacturing process, and given ARM's dominance in the embedded market, it's highly likely that other low cost Chinese chip outfits are not far behind.
While Intel gets ready to take on ARM, the British chip designer appears to be recruiting chipmakers to produce products that will encroach into Intel's back garden. µ
I would seriously do your homework before making statements such as " Arm is only good for phones" and " Great.... a faster smartphone chip".
In reality, Linux runs remarkably well on arm architecture, and in the server market, i see amazing advantages for having a dual core / quad core setup with arm running at its core. Imagine a file or web server only utilizing 8 watts. Amazing cost savings !!
Well this certainly opens a new front in the battle vs Intel.<yikes Proves Moore IS Better. But don't count out Tegra or Broadcom. I wouldn't bet on Intel not doing something revolutionary, maybe CPU+GPU+Memory on a single chip. People run Apps not architectures.
Intel can't get Meego off the ground, much less a more desktop/server oriented option. Locking Microsoft out would be foolish, even if it were to gain a bargaining position to keep Windows off ARM cpus, which again, won't happen. Microsoft has the advantage in that all they have to do is make their OS run on as many chips as possible. No small task but easier than convincing current ARM customers to step back to x86...just because.
Acquiring an ARM license might be a combination of insurance in the event that Intel decides to close their garden to a proprietary OS, as well as a natural evolution on their gaming console if they decide this particular hardware is suitable for that particular niche. In which case, nVidia might be a candidate for a future takeover.
Windows 8 will most likely support ARM and some form of efficient realtime x86 emulation isn't far off. Ludicrous? I think not. More likely a natural progression away from proprietary instruction sets. Microsoft has their own ARM license now, Nvidia is working on x86 emulation on tegra/gpus with former Transmeta tech, AMD's Bobcat will destroy Atom for a cheaper price and if Fusion takes off, Intel has no answer on the graphics side as Larrabee was a bust. Not to mention the fact that few if any companies want x86 in their smart phones. A buy out of ARM won't be allowed so Intel can't buy themselves into the mobile market either. Ludicrous? I think hilarious would be a better description.
BS indeed - Linux is quite well supported on ARM, and I bet Microsoft are busily working away at a port of Windows (why else did they just get an ARM foundry license???).
The whole IT world could be radically reshaped by the rise of ARM. The first data centre guys who can make ARM work for them are probably staring at a massive power bill reduction. The guys who stick to x86 are in severe danger of going bust.
Waiting for Microsoft to do a port might not be quick enough either - Linux operators could just waltz ahead right now.
Of course, Intel might make Atom comparable to ARM on the power budget side, but they'd better hurry up.
Quick guestimates:
ARM A15: 16 cores, 2.5GHz, say 8Watts.
Intel something-or-other: 4 cores, 3ish GHz, 120Watts + chipset costs.
Whichever way you look at that, I can't see an Intel chip delivering 15 times the performance for 15 times the power consumption, especially if your work load is actually IO bound not compute bound. Maybe the touchstone will be to see what Google do.
This is of course complete BS, there is enough OSS software used in business which could be used on ARM easily.
The iPad shows there is demand for bigger formfactors with a different architecture.
As an embedded systems engineer I would be very interested to use chips like this in upcoming designs for consumer devices.
Whatever the ARM chip is: what it is not is x86 compatible.
The business world runs on x86. Running x86 emulation on ARM is ludricrous.
Great so we have a smart phone cpu that is very fast!
Bring it up to speed against the depth of x86 microarchitecture and gpu performance and the power curve die size also increases.
That would be nice to have!