Unobtrusive Real-Time Tracing of Parallel Programs
MetadataShow full metadata
The US loses approximately $20 to $60 billion a year due to software bugs and glitches, many of which could be avoided if software developers had perfect visibility of the system state during program execution. However, achieving complete visibility of all internal signals in real time in modern systems is not feasible due to limited I/O bandwidth, high internal complexity, and high operating frequencies. This thesis aims to develop a method that can be implemented in hardware to enable unobtrusive real-time tracing. The proposed method exploits several components (e.g., predictors, reductors, etc.) to reduce the size of the trace information through a tree-shape multiple level algorithm. The results show that with this method traces can be compressed with a relatively high ratios.