Utpal Banerjee’s scientific contributions

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (5)


Fast Execution of Loops with IF Statements
  • Article

December 1984

·

7 Reads

·

12 Citations

IEEE Transactions on Computers

Utpal Banerjee

·

A parallel method of execution for a certain class of loops containing IF statements is described. We replace a given loop by an equivalent set of five loops, four of which are vectorizable; the fifth loop is executed in hardware as a Boolean recurrence. The proposed architecture handles all loops that produce recurrences with order ≤m, a hardware parameter.


Fast Execution of Loops with IF Statements.

June 1984

·

5 Reads

·

2 Citations

ACM SIGARCH Computer Architecture News

In this paper we show how to execute in parallel loops containing IF statements. We give an architectural model of parallel computation and describe the design of a hardware Boolean Recurrence Solver. Our method of handling such loops is then compared with those used by some of the existing supercomputers.


Fast execution of loops with if statements

June 1984

·

7 Reads

ACM SIGARCH Computer Architecture News

In this paper we show how to execute in parallel loops containing IF statements. We give an architectural model of parallel computation and describe the design of a hardware Boolean Recurrence Solver. Our method of handling such loops is then compared with those used by some of the existing supercomputers.


Accessing sparse arrays in parallel memories

January 1983

·

11 Reads

·

6 Citations

Journal of VLSI and Computer Systems

The concept of dense and sparse execution of arrays is introduced. Arrays themselves can be stored in a dense or sparse manner in a parallel memory with m memory modules. The paper proposes hardware for speeding up the execution of array operations of the form c(câ+ci)=a(aâ+ai) op b(bâ+bi), where aâ, a, bâ, b, câ, c are integer constants and i is an index variable. The hardware handles 'sparse execution', in which the operation op is not executed for every value of i. The hardware also makes provision for 'sparse storage', in which memory space is not provided for every array element. It is shown how to access array elements of the above form without conflict in an efficient way. The efficiency is obtained by using some specialised units which are basically smart memories with priority detection, one's counting or associative searching. Generalisation to multidimensional arrays is shown possible under restrictions defined in the paper. 12 references.


ARRAY MACHINE CONTROL UNITS FOR LOOPS CONTAINING IFs.

January 1980

·

7 Reads

·

3 Citations

The control unit is the interface between the compiler and the processing part of a computer. A number of array (parallel or pipeline) machines have been built with scalar or array instruction sets. Most such machines do a poor job of handling sparse data arrays, and this paper addresses how such computations may be better handled.

Citations (2)


... In this section we address the problem of run time resolution of data dependencies, by the use of speciel purpose hardware. Banerjec and Gajski [4] use speciel purpose hardware to speed up Loops with IF statements, but the example considered by them involves dependencies that cannot be resolved at all, and the hardware merely speeds up the sequential execution by replacing microcude by logic gates, and furthermore involves a possibly exponential increase in the amount of hardware required as the full condition tree must be evaluated in parallel. We are more interested in cases where the dependenciescannot be resolved at compile-time, but can be resolved at run time sufficiently early so that speedup is not constrained by the potential dependency. ...

Reference:

A hardware accelerator for speech recognition algorithms
Fast Execution of Loops with IF Statements.
  • Citing Article
  • June 1984

ACM SIGARCH Computer Architecture News

... With the addition of vector units (the Cray-1) or SIMD operation [4] compiler writers went wild trying to automatically parallelize high-level programs [15], with some success [21]. While there may have been some initial hope of parallelizing general purpose programs with complex control flow [3, 8], this was quickly dropped in favor of scientific programs with simpler control flow and, hopefully, " embarrassing " parallelism. As science and engineering of all kinds quickly progressed, it became apparent that the relevant computations required were not all that straightforward. ...

Fast Execution of Loops with IF Statements
  • Citing Article
  • December 1984

IEEE Transactions on Computers