Unisys Corporation

Steven T Hurlock

Robert A Johnson

Andrew Ward Beale

Arthur J Nilson

Patricia Nichols

Stanley P Naddeo

Stephen Kun

Jeremy S Nichols

Systems and methods for validating operation of a compiler are disclosed. One method includes receiving a definition of language syntax at an automated source code generator, and generating program code at the automated source code generator, the program code represented in source code including constructed self-validating code and syntactically-correct automatically generated code. The method also includes providing the source code to the compiler to be compiled into object code. Based on execution of the self-validating code as object code in a computing system, a computing system executing the object code outputs an indication of the correctness of compilation of the program code.

Syntax language generator for compiler validation

Code coverage questions are addressed by a code coverage method that instruments an electronic module source design file with coverage probes and gives hierarchical names to the probes, then provides therefrom an instrumented gate level netlist. The instrumented netlist is run on a hardware emulator, executing reset trigger scripts to reset the branch and statement probes, and then a fully initialized design is driven in emulation on a simulated testbench from which the probe values are retrieved. These values can then be evaluated to determine the extent of code coverage. Various forms of coverage are supported including branch, statement, reset trigger and toggle coverage.

Code coverage testing in hardware emulation

A code coverage tool provides that a netlist is instrumented with gates for providing a comparison of an output of the design gates on one cycle with their output on a next cycle to determine if the gate was exercised during an emulation.

Finding unexercised logic for use in code coverage testing

A code coverage tool can be made more useful if circuits of a specified function can be identified by their truth table. This knowledge can be used during code coverage testing to determine what circuits may not be covered during the appearance of specified input data configurations for such circuits. Forcing all circuits to a specified state and running the truth table data through them, when properly instrumented in emulation allows for determination of such specified functionality.

Identifying specific netlist gates for use in code coverage testing

A code evaluation algorithm is used to determine optimal cache parameters. In evaluating the code, the algorithm uses a set of code characteristics the presence of which, or the degree of presence of which, impact the proper selection of cache parameters. Once the code characteristics have been selected, the algorithm uses boolean truth state tables or fuzzy logic membership set definitions to specify whether and when each code characteristic is true or false (i.e., present or absent) for a given set of code instructions. Programs are classified as having the following characteristics: 1) arithmetic intensive; 2) logic/decision intensive; 3) reference intensive; and/or 4) array/vector/table processing. To evaluate the code, the degree of presence or absence of these characteristics is described in logical fashion, using fuzzy logic. To logically evaluate and relate such &ldquo;fuzzy&rdquo; concepts, the algorithm uses fuzzy logic membership functions to determine the degree to which there exist a small number, medium number or large number of instances of a particular class of operator in the code analyzed. Next, the algorithm defines a fuzzy logic rule base which relates (e.g., by implication) the presence or absence (i.e., truth or falsehood) of the condition impacting proper selection of cache parameters to conclusions regarding the selection of cache parameters. The algorithm then determines (evaluates) the conclusions regarding the proper selection of cache parameters by evaluating the conditions impacting the conclusions while applying the rule base rules. In particular, the algorithm uses the &ldquo;Mandani Product&rdquo; inference engine for computational simplicity. The center of gravity of the fuzzy outcomes is computed and the result is discretized yielding a crisp, meaningful result representing an optimal cache parameter choice.

Code type control of caching parameters

In a data processing system comprising a first level cache, a second level cache, and a processor return path, wherein only one of the first level cache and second level cache can control the processor return path at a given time, an improvement comprises a queue disposed between an output of the first level cache and the processor return path to buffer data output from the first level cache so that the first level cache can continue to process memory requests even though the second level cache has control of the processor return path.

Growing community of inventors

Wynnewood, PA, United States of America

Steven T Hurlock