Setting its sights on audio and voice applications, the second-generation TeakLite-II DSP core and audio subsystem clocks at speeds of up to 200 MHz. It's based on the TeakLite DSP core, which contains a mix of DSP and control instructions.
Developed by CEVA-DSP, TeakLite-II requires just 0.5 mm2 when implemented with 130-nm design rules. The small core area and high code density, which are possible thanks to the 16-bit instruction set, keeps system costs low. The subsystem also minimizes power, as the core consumes just 0.1 mA/MHz when running a stereo MP3 decoding algorithm.
The subsystem peripherals that will optimize implementation for audio applications surround the basic DSP core. These peripherals include a cache memory, configurable data and program memories, a three-channel DMA controller, and external interfaces. They add another 1.5 mm2 to the core area, bringing the total DSP subsystem to 2 mm2.
The external interface options include an ARM high-speed bus (AHB) master interface and an ARM peripheral bus (APB) interface to leverage already designed APB peripherals such as the power-management unit, timers, general-purpose I/Os, and interrupt controller.
The data memory can be configured as 15, 24, or 32 kbytes, while the program memory can be set at 4 or 16 kbytes. Designers can add a tightly coupled memory of up to 4 kbytes to handle time-critical algorithms. Options include interfaces for audio codecs and time-division multiplexed devices.
The combination of the DSP core and peripherals provides a good match to handle audio algorithms from all over the world—MP3, WMA, AAC, HE-AAC/aacPlus, Ogg Vorbis, BSAC, AC3, and DTS. The core readily handles speech/voice codecs such as G.7231.1, G.729, G.726, G.711, G.168, AMR, and WB-AMR. An integrated development environment that extends the CEVA-Audio tools available for previous TeakLite DSP cores lets designers easily develop and debug their applications.
The core/subsystem is now available for licensing. Contact the company for terms.