Using warp as a supercomputer in signal processing

The Warp Computer: Architecture, Implementation, and Performance

“…Since the two copies of the wire-wrapped prototype Warp machine became operational at Carnegie Mellon in 1986, we have used the machines substantially in various applications [2,3,4,10,13,22]. The application effort has been increased since April 1987 when the first PC Warp machine was delivered to Carnegie Mellon.…”

“…Many of the algorithms on Warp make use of it, including most of the low-level vision programs, the discrete cosine transform (DCT), singular value decomposition [2], connected component labeling [22], border following, and the convex hull. The last three algorithms mentioned also transmit information in other ways; for example, connected components labeling first partitions the image by rows among the cells, labels each cell's portion separately, and then combines the labels from different portions to create a global labeling.…”

Research on Signal Processing Supercomputers

“…" 512 x 32 The first 10-cell prototype was completed in February 1986; delivery of production machines by our industrial partner (GE) started in April 1987. Extensive experimentation with both the prototype and production machines has demonstrated that the Warp architecture is effective in the application domain of robot navigation, as well as in other fields such as signal processing, scientific computation, and computer vision research [2,1,3,5,6,141. For these applications, Warp is typically several hundred times faster than a VAX 11/780 class computer.…”

“…as depicted in Figure 1. We dcsc'1hC this machine only briefly here: inore detail is aaiIlable separately [2]. The Warp processor array performs the In addition to generating addresses, the interface unit passes data and results between the host and the Warp array, possibly performing some data conversion in the process.…”

Computational models for parallel computers

“…Then during computation, entries of matrix A will be input to the left-most cell in the row-major ordering, and shifted to the right from cell to cell, as depicted in Figure 6 There are many other usage examples based on the local computation model. They include the discrete cosine transform [2] and the labeled histogram computation [17].…”