Sent to you via Google Reader
At Mobile World Congress, NVIDIA officially took the wraps off the 1.5GHz, quad-core Tegra 3 that had been leaked early last month.
NVIDIA isn't shy about making bold claims for its upcoming chip, starting with the codename: Kal-El (Superman's Kryptonian name). The company also showed the 40nm chip beating a Core 2 Duo (T7200) in Coremark, which is impressive, but mainly serves to highlight how shameful it is that certain computer makers still ship "premium" machines with such a creaky old part in them.
The launch timeframe for Kal-El is aggressive, with NVIDIA promising tablets based on the new part by August of this year, and smartphones following at the end of the year. The chip will also support display resolutions even higher than what was specified in the leaked slides, with NVIDIA demoing it pushing a 2560x1600 screen. On the media processing side, Anandtech is reporting that Kal-El finally brings support for NEON—the ARM answer to Intel's SSE vector instructions—to the Tegra line. There's also a 12-core GPU, but, like its Tegra 2 predecessor, it won't be a unified shader architecture design. Feeding that media hardware is a 32-bit LPDDR2 controller, which means a bit less bandwidth than we'd have expected, but it's probably a wise choice given power considerations.
This chip will make a great mobile gaming part, and we certainly hope that there's some truth to the rumors of Sony using it in an upcoming media/gaming tablet. It could also end up running more than just PlayStation games, though. On the most recent earnings call, NVIDIA CEO Jen-Hsuan referenced the recently unveiled Windows ARM port and suggested that the Tegra line could be making its way into "PC" notebooks.
We envision Tegra bringing us to a future, Bizarro-World moment in which Microsoft's desktop OS runs on a classic RISC ISA while Apple's desktop OS is confined to x86. On that day, we expect to read a slew of confident editorials at Mac sites about how RISC is old and bust...