Re: Interesting numbers of note......And now China $5.-
in response to
by
posted on
Jan 19, 2015 01:56AM
The $5, 64-bit Processor and Disruption Ahead
Peter Clarke
07 January 2015
http://electronics360.globalspec.com/article/4874/the-5-64-bit-processor-and-disruption-ahead
Several shivers may have run up the backs of executives at Intel and some ARM licensees when they heard that fabless chip company Allwinner Technology Co. Ltd. (Zhuhai, China), has not only pushed on to offer a 64-bit system-chip for tablet computers but, more significantly, is pricing it at just $5.
At that price the system-chip, while still the processing heart of the tablet, will become one of the lower cost components in the assembly, which is not what Intel or other ARM licensees want to hear.It's not necessarily what ARM wants to hear either because even though multiple cores on an SoC give it a greater royalty it is as a percentage of the chip price. And a bigger percentage of $5 is not much at all.
In the era of the $5 64-bit processor, the packaging the tablet computer comes in will likely cost more than this Allwinner SoC and, given the importance of the consumer unboxing experience, may be more important to the tablet's final success in the market.
Tablet specialist
Allwinner has been a tablet computer specialist but has mainly pitched its devices for Chinese branded Android tablets that are priced to drive sales in China. The company moved up to 32-bit octa-core designs in the form of the A80 and the A83T in 2014 both produced on TSMC's 28HPM 28nm CMOS manufacturing process.
Now, a little later than competitors such as Apple, Qualcomm and Intel, Allwinner is moving to 64-bits for tablet applications, with the rather unimaginatively named A64. The move was expected (see Allwinner Preps 64-bit Tablet Processor) but the price point may come as a surprise. And the timing is right because prior to the introduction of the Android 5.0 Lollipop operating system Android devices could not make much use of 64-bit processing.
Despite the fact that Linux – which is the foundation of Android – has supported 64-bit operation, the Android kernel and higher software layers have all been written on the supposition of 32-bit working. Now with Android version 5.0 the Dalvik virtual machine is being replaced with ART or Android Runtime, which is a cross-platform runtime environment that fully supports x86, ARM and MIPS processor architectures in both 32- and 64-bit forms.
Allwinner claims that priced at $5, the A64 is the most cost-efficient 64-bit tablet processor yet produced. The design is a quad-core Cortex-A53 design supporting H.265/H.264 decoding in hardware and HDMI 4K display and sports Allwinner's SmartColor display technology. It also supports the Trusted Firmware security architecture from ARM. The company said it will release Android 5.0 on the A64 and that it will support the latest Linux kernel version 3.10 to match the 64-bit architecture and the Android 5.0 operating system.
It is not yet clear what Allwinner has done about graphics – it has designed both Imagination's PowerVR GPUs ARM's Mali GPUs into its SoCs in the past. It is also not clear whether the low price is the result of reusing IP cores on a stable 28nm manufacturing process, or from moving to 20nm planar CMOS and taking some benefit in reduced die area. A spokesperson for Allwinner said more detail will be disclosed soon.
At $5 the A64 will no doubt continue to help drive tablet computers for the emerging world down in price. But if a quad-core 64-bit A64 with decent graphics can be priced at $5 then a 64-bit octa-core at a similar single-digit dollar price is certainly possible. And that could likely meet the needs of an Android Lollipop tablet destined for any where in the world.
And at that price would Intel, Qualcomm or Samsung wish to compete? And where the processor at the heart of the Android tablet processor goes, the smartphone SoC and PC processors are likely to follow.
The lesson is that the value is moving away from the plain old CPU and towards more specialized peripheral capabilities, such as sensor fusion, vision processing and systems that learn, and to the system architectures that encompass them.