• Corpus ID: 32756913

The MIT database accelerator: 2K-TRIT circuit design

@article{Wade1987TheMD,
  title={The MIT database accelerator: 2K-TRIT circuit design},
  author={J. P. Wade and Peter J. Osler and Richard Zippel and Charles G. Sodini},
  journal={1987 Symposium on VLSI Circuits},
  year={1987},
  pages={39-40}
}
The computational speed of a conventional von Neumann computer architecture is limited primarily by the data path bottleneck between the memory and the central processing unit (CPU). One solution to this problem is to eliminate the separation between the CPU and memory by moving the processor functions directly into the memory. The Database Accelerator (DBA) chip is an attempt to merge these two functions by incorporating high density content-addressable memory (CAM) in a multi-purpose parallel… 
5 Citations

Figures from this paper

Programming the data structure accelerator
  • R. Zippel
  • Computer Science
    Proceedings of the 5th Jerusalem Conference on Information Technology, 1990. 'Next Decade in Information Technology'
  • 1990
A fine-grained, massively parallel SIMD (single-instruction-stream, multiple-data-stream) architecture, called the data structure accelerator, is presented, and its use in a number of problems in
A 20 kbit associative memory LSI for artificial intelligence machines
TLDR
The associative memory, with its highly efficient associative operation capabilities, promises to be a large step toward the development of high-performance artificial intelligence machines.
Content-Addressable Memories for Quadtree-Based Images
TLDR
The authors have reviewed a range of quadtree processing functions relevant to computer graphics and Image processing, and some new algorithms have been discovere, and the proposed VLSI chip has microcoded logic on each row, as well as its CAM cells.
A parallel string search algorithm
TLDR
A new parallel processing algorithm for solving string search problems is presented that requires two and only two iteration steps to find the pattern in the text, while the best existing parallel algorithm needs the computation time O(loglog n).
Content-addressable memory for VLSI pattern inspection
TLDR
A custom VLSI chip was designed with a 2- mu m double-metal CMOS technology, which can perform parallel template matching at the rate of 1.6*10/sup 7/ local images per second.

References

SHOWING 1-2 OF 2 REFERENCES
Dynamic cross-coupled bitline content addressable memory cell for high density arrays
This paper describes the design of a novel dynamic Content Addressable Memory (CAM) cell suitable for high density arrays on the order of 64K bits. The proposed cell is capable of storing three
The MIT Database Accelerator: A Novel Content Addressable Memory
  • Wescon Digest 01 Technical Papers, 1986.
  • 1986