LONDON The A6 processor inside Apple's iPhone 5 mobile phone is a dual-core Cortex-A15 manufactured for Apple by Samsung Electronics Co. Ltd. in its 32-nm HKMG manufacturing process, according to analysts at Nomura Equity Research.
This would mean Apple is one of the first companies to introduce a Cortex-A15-based processor. Cortex-A15 is the highest performance processor core from intellectual property licensor ARM Holdings plc (Cambridge, England).
[ARM TechCon 2012, the largest ARM design ecosystem under one roof, is Oct. 30 - Nov. 1 in Santa Clara. Click here to learn more]
Samsung said it had started sampling the industry's first dual-core ARM Cortex-A15 processor late in 2011, the Exynos 5250, made with its 32-nm HKMG process and intended for volume shipment in summer 2012. The Exynos 5250 includes Mali graphics, and is intended for use in high-end tablet computers. Its 2-GHz clock frequency is claimed to double the performance of the previous 1.5-GHz dual-core Cortex-A9 based Exynos.
Apple provided few details when it launched the iPhone 5 on Wednesday (Sept. 12) about the smartphone's application processor and graphics capability. The company did say the A6 processor provided twice the CPU performance and twice the graphics performance of the A5x used in the iPhone 4S.
Nomura provided no source for its report nor a clock frequency for the processor. Typically, mobile phone application processors run with clock signals of up to 1.5 GHz. However, designing in the Cortex-A15 could help explain how Apple has achieved performance equivalentto the iPhone 4S
Apple is expected to retain graphics IP licensor Imagination Technologies Group for the graphics rendering portion of the chip. The Apple A5 processor is reported to use the dual-core PowerVR SGX543MP2, so the A6 could use the quad-core version, the PowerVR SGX543MP4.
Taiwan Semiconductor Manufacturing Co. (Hsinchu, Taiwan) was reportedly working on a version of the A6 processor for Apple in 2011. It was rumored then to be a quad-core design for implementation in 28-nm manufacturing process and was expected to debut in the third-generation iPad.
If Samsung is the sole supplier of the A6 processor as indicated by Nomura analysts this squares with recent predictions that TSMC is working on pulling in its 20-nm process and working to supply Apple in the second-half of 2013 using that process.
"the use of Cortex-A15 could help explain how Apple has achieved the equivalent talk time to the previous iPhone 4S" - no it couldn't. The die shrink from 45nm to 32nm would explain that.
Why does EETimes keep publishing baseless speculation from unqualified analysts? The analysts in that iphone 5 predictions article yesterday got 2 out of 9 predictions correct.
there is a pro-ARM ( pro Brit ) agenda being pushed here by the reporter based in Londonistan at the expense of giving US readers of EE Times a well-informed or balanced perspective. Thx to EE Times employment policies we will just have to do it ourselves.
This should be an interesting mix of processing power vs battery life. If the performance has been jacked up to 2x as they claim then assuming it consumes the same power as it's predecessor it should still result in longer battery life and faster response times.
So I guess the quad core version of it will go into next Ipad.
And dont forget there is the chance to upgrade with a couple of Cortex-A7 cores tucked in to a corner of a dual-core Cortex-A15 processor design.
That would create a quad-core processor that implements the "big-little" power-saving technique.
Such chips are expected to arrive and be powering smartphones in 2013 as I remember Warren East saying at the launch of Cortex-A7 in 2011.
Well in my opinion it is too early to see an A15-based SoC powering such an important and high volume product as the iPhone.
The 2x claim by Apple is quite unsubstantial and not based on a specific metric (e.g. a benchmark). Therefore compared to the A5 processor in iPhone 4S, my best guesses are that this "2x claim" could come from the following:
- using 32nm process instead of 45nm could lead to an increase from 1GHz dual A9 to 1.5 GHz (maybe more?) again dual A9
- certain tweaks here and there
- beefier memory subsystem (maybe another memory controller or higher frequency DRAM?)
- beefier graphics
- software improvements.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for todays commercial processor giants such as Intel, ARM and Imagination Technologies.