A view of Fermi’s 3 billion+ transistors.
Nvidia brought two major product chipsets to CES: the Tegra CPU (for netbooks and tablet computers) and its new top of the line GPU, the GF100, nicknamed Fermi.
Nvidia’s Tegra 2 processor goes up directly against Intel’s Atom, which comes in just about every netbook out there it seems. Just now getting on the market, the Tegra 2 is based on the well-regarded Arm processor core, long valued for its low power chops.
Nvidia targets the chip at devices with 5-inch to 15-inch screens; the dual-core processor can play back 1080p high-definition video while conserving battery life, an advantage over Intel’s product. Innovative Converged Devices, Asus, and T-Mobile were among companies showing concept products built around the Tegra. Lenovo used another Arm-based CPU, from Qualcomm, for its lightweight, Linux-based Skylight smartbook.
For anyone involved in graphics and animation—or if you’re an avid gamer—Nvidia’s Fermi GF100 was the draw. Naming it to honor the famed physicist Enrico Fermi is more than just a cute marketing ploy: besides offering high-performance ray tracing, the massive chip (over 3 billion transistors, compared to the previous generation GPU’s 1.4 billion) has 512 CUDA cores and other specific circuitry to boost physics, finite element analysis, and other high-precision scientific computing tasks. Nvidia offers an interesting series of overviews on Fermi here.
The Fermi chip, expected to release by March, supports Microsoft’s new DirectX 11 architecture, offers hardware support for boosting GPU voltage and other tricks beloved of overclockers, and provides a complete hardware decode for 3D-based Blu-ray.
Quick overview video on Fermi
Nvidia’s CES blog
Check out the chip’s capability of generating real time physical effects in games in this video of a demo.
For those who want more detail, here’s a good tech overview.
