A shared memory module for asynchronous arrays of processors

Authors:
Michael J. Meeuwsen;Zhiyi Yu;Bevan M. Baas
Affiliations:
Department of Electrical and Computer Engineering, University of California, Davis, CA;Department of Electrical and Computer Engineering, University of California, Davis, CA;Department of Electrical and Computer Engineering, University of California, Davis, CA
Venue:
EURASIP Journal on Embedded Systems
Year:
2007

Citing 5
Cited 2

Smart Memories: a modular reconfigurable architecture

Proceedings of the 27th annual international symposium on Computer architecture
On-chip vs. off-chip memory: the data partitioning problem in embedded processor-based systems

ACM Transactions on Design Automation of Electronic Systems (TODAES)
A Case for Intelligent RAM

IEEE Micro
Scratchpad memory: design alternative for cache on-chip memory in embedded systems

Proceedings of the tenth international symposium on Hardware/software codesign
Globally-asynchronous locally-synchronous systems (performance, reliability, digital)

Globally-asynchronous locally-synchronous systems (performance, reliability, digital)

Programming Reconfigurable Decoupled Application Control Accelerator for Mobile Systems

ARC '08 Proceedings of the 4th international workshop on Reconfigurable Computing: Architectures, Tools and Applications
Coordinated concurrent memory accesses on a reconfigurable multimedia accelerator

Microprocessors & Microsystems

Quantified Score

Hi-index	0.00

Visualization

Abstract

A shared memory module connecting multiple independently clocked processors is presented. The memory module itself is independently clocked, supports hardware address generation, mutual exclusion, and multiple addressing modes. The architecture supports independent address generation and data generation/consumption by different processors which increases efficiency and simplifies programming for many embedded and DSP tasks. Simultaneous access by different processors is arbitrated using a least-recently-serviced priority scheme. Simulations show high throughputs over a variety of memory loads. A standard cell implementation shares an 8 K-word SRAM among four processors, and can support a 64 K-word SRAM with no additional changes. It cycles at 555MHz and occupies 1.2mm2 in 0.18 µm CMOS.