Ay. Wu et al., SYSTEM ARCHITECTURE OF AN ADAPTIVE RECONFIGURABLE DSP COMPUTING ENGINE, IEEE transactions on circuits and systems for video technology, 8(1), 1998, pp. 54-73
Modern digital signal processing (DSP) applications call for computati
onally intensive data processing at very high data rates. In order to
meet the high-performance/low-cost constraints, the state-of-the-art v
ideo processor should be a programmable design which performs various
tasks in video applications without sacrificing the computational powe
r and the manufacturing cost in exchange for such flexibility. Current
ly, general-purpose programmable DSP processor and application-specifi
c integrated circuit (ASIC) design are the two major approaches for da
ta processing in practical implementations. In order to meet the high-
speed/low-cost constraint, it is desirable to have a programmable desi
gn that has the flexibility of the general-purpose DSP processor while
the computational power is similar to ASIC designs. In this paper, we
present the system architecture of an adaptive reconfigurable DSP com
puting engine for numerically intensive front-end audio/video communic
ations. The proposed system is a massively parallel architecture that
is capable of performing most low-level computationally intensive data
processing including finite impulse response/infinite impulse respons
e (FIR/IIR) filtering, subband filtering, discrete orthogonal transfor
ms (DT), adaptive filtering, and motion estimation for the host proces
sor in DSP applications. Since the properties of each programmed funct
ion such as parallelism and pipelinability have been fully exploited i
n this design, the computational speed of this computing engine can be
as fast as ASIC designs that are optimized for individual specific ap
plications. We also show that the system can be easily configured to p
erform multirate FIR/IIR/DT operations at negligible hardware overhead
. Since the processing elements are operated at half of the input data
rate, we are able to double the processing speed on-the-fly based on
the same system architecture without using high-speed/full-custom circ
uits. The programmable/high-speed features of the proposed design make
it very suitable for cost-effective video-rate DSP applications.