Electronic Design

New Transistors, Less Power Mark 2007

Digital technology saw some significant leaps this year as companies responded to the limits of Moore’s Law and the need for better power conservation.

For more than 40 years, “traditional” transistors have been built using a combination of polysilicon- gate electrodes and silicon-dioxide (SiO2) dielectric insulators because of the materials’ manufacturability and ability to deliver continued transistor performance improvements.

Yet with Moore’s Law threatening to expire as silicon performance reaches its physical limitations, companies like Intel and IBM have introduced hafnium-gate dielectric insulators and new non-disclosed metal materials for the NMOS and PMOS transistors that make up CMOS semiconductors (Fig. 1). Intel is leading this change with the first working 45-nm multicore processors using these technologies. The company claims that “these new materials, along with the right process recipe, reduce gate leakage more than 100-fold while delivering record transistor performance.”

On Nov. 12, Intel launched the Penryn family of multicore processors, which will run Windows Vista, Mac OS X, and Linux operating systems. The Penryn architecture will be used to build 35 new models of CPUs for both servers and high-end PCs.

Not all companies can afford to partner with the likes of IBM to access new hafniumbased process technology. Therefore, companies like Mosaid Technologies offer a cheaper alternative. Mosaid’s Mobilize IP platform addresses key leakage issues at 90 nm and below without requiring changes to the silicon process.

The new IP manages leakage issues by providing both static and active power management in the form of a configurable standard-cell library (Fig. 2). The library is used to form voltage- and frequency- scalable power islands.

An automated hardware- or softwarecontrolled power-island manager (PIM) administers all power-island sequencing of active and sleep phases over process, voltage, and temperature. This eases the instantiation of power islands while reducing leakage over 100-fold. It also can be used in conjunction with 1-V and 0.8-V supplies. And, it can transition to and from sleep mode in a mere 50 ns.

High-performance mobile audio/video (A/V) chips often deliver superior processing at the cost of higher power–and lower power means less processing muscle. The SVENm (Scalable Video ENgine) from On Demand Microelectronics, though, has changed that paradigm.

Using as little as 80 mW for SD decode and 300 mW for encode, this chip can handle today’s compute-intensive video standards, like H.264, MPEG-4, and VC-1. The SVENm targets mobile applications that require video formats up to D1 (720-by-576 PAL or 720-by-480 NTSC) resolution. It also excels where most processors fail by delivering the ability to handle future video standards via software programmability.

The individual functional blocks are programmable using C and integrated development tools. The SVENm’s processor architecture was built based on an in-depth analysis of video-processing algorithms, yielding an optimized processor architecture. The architecture exploits parallelism to process video, resulting in increased efficiency.

This feature set makes the SVENm optimal for use in smart phones and portable media devices, enabling designers to add drop-in video capability to their next-generation MP3 players. Other features include simultaneous video encoding and dual-stream decoding, audio processing, and high-performance image processing.

TAGS: Intel
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.