Samsung confirmed to use ARM's big.LITTLE chip architecture for frugal Exynos in 2012
At the ARM TechCon 2011 event we got the much desired confirmation by Samsung's VP John Kalkman, that the mobile device maker will use the new Cortex-A7 plus ARM's latest big.LITTLE architecture to build a new Exynos chip in 2012: “I’m extremely excited to announce that Samsung will deliver a new Exynos processor in 2012 that leverages both the Cortex-A7 and big.Little technology to meet the crucial demands of always-on and always-connected computing.”
The new chip paradigm does what some chipmakers like NVIDIA or TI have been doing with their multicores for a while, combining low-powered cores with high-performance ones to achieve unmatched battery vs performance ratio. ARM leverages that paradigm for its new frugal Cortex-A7 chip, which has the design of Cortex-A15, but with less oomph, and combines it with actual A15 cores for big.LITTLE, allowing racing performance when needed, and power-sipping operation during everyday tasks, making the whole system fast, but frugal and more affordable. Here's a funny cartoonish demo of the concept by ARM:
This news makes us jump from joy just imagining the future phones from Samsung - finally a design which is built from the ground up with the new requirements towards battery life on mobile devices. The big.LITTLE Exynos is supposed to come in the second half of 2012, although they most likely mean sampling at that time, since ARM hinted at phones with the platform not appearing until 2013. This means that the Samsung GALAXY S III will still feature the quad-core Exynos rumored, which, due to the die shrinkage should be more powerful than today's dual-core one, and at the same time more efficient. Texas Instruments has also licensed Cortex-A7 and big.LITTLE, so after TI OMAP5 we are expecting a similar chip from the Texans, not to mention they are a reference platform for Android ICS now.
In short, we can expect the performance wars to peak next year with the quad-core chippery, and the focus afterwards to shift on the power consumption/performance/affordability ratio. Not much software nowadays that can squeeze multicore processors for what they're worth anyway.