A machine method for square-root computation

Bemer, R. W.

doi:10.1145/368685.368690

Cited by 7 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Iterative methods that require only programmable shift, addition/subtraction, selection, and table look-up operations are an alternative to MAC-based methods [16]- [21]. Existing reconfigurable arrays either do not provide architectural support for these operations or embed dedicated shift units in the reconfigurable fabric.…”

Section: Dsp-oriented Reconfigurable Arraysmentioning

confidence: 99%

See 1 more Smart Citation

Reconfigurable array for transcendental functions calculation

Sima

McGuire

Miller

2008

2008 International Conference on Field-Programmable Technology

View full text Add to dashboard Cite

Expanding transcendental functions in a series of Shiftand-Add operations is an alternative to Taylor or Chebyshev series expansions when fixed-point arithmetic with reduced wordlength is required. Typically, reconfigurable arrays do not provide architectural support for shift operations. Instead, shift operations are emulated by either multiplexing logic or multiplication by a power of 2. In this paper we describe the architecture of a reconfigurable array that can natively support shift operations. Rather than augmenting the reconfigurable fabric with dedicated shift units, the interconnection network is extended with shift capabilities. This is conceptually possible since a shift operation is a rearrangement and not a combination of the signals. Layers of computing tiles supporting Shift-and-Add/Subtract and Add-and-Select operations are interleaved with interconnect layers. On such a reconfigurable array, a variety of transcendental functions can be efficiently implemented.

show abstract

Section: Dsp-oriented Reconfigurable Arraysmentioning

confidence: 99%

“…. M + ∆) [16]. In parallel, by cyclic multiplication by 1.0 or the factors A i , the square root, √ m k , will fall within the range ( √ M .…”

Section: Calculation Of Square Rootmentioning

confidence: 99%