Verifying GPU Kernels by Test Amplification

The authors present a novel technique for verifying properties of data parallel GPU programs via test amplification. The key insight behind their work is that they can use the technique of static information flow to amplify the result of a single test execution over the set of all inputs and inter-leavings that affect the property being verified. They empirically demonstrate the effectiveness of test amplification for verifying race-freedom and determinism over a large number of standard GPU kernels, by showing that the result of verifying a single dynamic execution can be amplified over the massive space of possible data inputs and thread inter-leavings.

Provided by: Association for Computing Machinery Topic: Software Date Added: Jun 2012 Format: PDF

Find By Topic