The prospect of quad-core or higher smartphones excites many, but it also evokes fear. Many already suffering from somewhat poor battery life so imagine such high core-count designs as being Shiva incarnate for smartphone batteries, a destroyer of your pleasant smartphone world.
But such fears may yet prove unfounded, as many manufacturers are cooking up unique solutions to having a high core count, while keeping a relatively lean power budget.
NVIDIA Corp. (NVDA), makers of the Tegra series of ARM system-on-a-chip (SoC) CPU/GPU designs today unveiled the secret of its plans to make its upcoming Kal-El quad-core SoC more power efficient. The secret is basically this -- NVIDIA's quad-core chip is really a penta-core.
Yes, Tegra 3 has a low-power fifth "companion" core that's essentially the kind of modest ARM design that power the superb power efficiency of devices like the original iPhone. NVIDIA writes [press release]:
During less power-hungry tasks like web reading, music playback and video playback, Kal-El completely powers down its four performance-tuned cores and instead uses its fifth companion core. For higher performance tasks, Kal-El disables its companion core and turns on its four performance cores, one at a time, as the work load increases.
NVIDIA has published two white papers on the technology and related Kal-El developments.
The incremental approach to enabling the cores means that Kal-El chips, which will likely be clocked up to 1.5 GHz and may deliver as good or better battery life as current models (providing you aren't doing video encoding and playing 3D games all day). In fact, NVIDIA says Kal-El delivers better battery life than its current generation Tegra 2 (dual-core) chips. In benchmarks, the new chips were also shown posting almost twice the benchmark score of rival Qualcomm Inc.'s (QCOM) MSM8660 dual-core SoC and Texas Instruments Inc.'s (TXN) OMAP4 (1 GHz) processor.
Note, Tegra 2 and Tegra 3 are both 40 nm chips produced by Taiwan Semiconductor Manufacturing Comp., Ltd. (TPE:2330) so NVIDIA isn't "cheating" with die shrinks on that test. However, Qualcomm uses TSMC's 45 nm technology in the MSM8660, while TI uses its own 45 nm process in the OMAP4 chips, thus the transistor size is slightly larger, so be aware that NVIDIA may not be comparing apples to apples, exactly, in its comparison with rival chip designs.
Kal-El tablets are now expected to be shipping this holiday season, in Q4 2011. Initially the tablets were slated for a Q3 2011 release (summer), but that date was pushed back. NVIDIA CEO Jen Hsun-Huang says that some of the delay is based on hardware partners "getting the industrial design as wonderful as possible, and some of it is related to tuning and performance."
For this holiday season the NVIDIA chip is expected to be found exclusively in Android tablets. However, Kal-El and the more powerful Wayne (Tegra 4) chips are expected to be featured prominently in ARM-based Windows 8 tablet designs next holiday season.
NVIDIA is packing a more powerful GPU core into Kal-El, which packs 12 streaming multiprocessors (SMPs). It says it plans to unveil more details about new features on the GPU side of the coin shortly.
The company claims to have a 70 percent market share among Android tablets and a little less than 50 percent market share in the smart phone market [source], though recent reports have indicated that NVIDIA still trails Qualcomm slightly in the smart phone market.
For those who don't read comics, Kal-El is the original name of DC Comics superhero Superman. While Tegra 3 and 4 are named after DC superheroes, Tegra 5-7 are codenamed after Marvel Comics heroes ("Grey", "Logan", and "Stark", respectively). Stark (Tegra 7) is expected to launch in 2014.