Looking Beyond Performance: Processors for Time Travel

Dr. Narayanasamy is from UCSD
The processor industry is at an inflection point. In the past,
performance was the driving force behind the processor industry.
But in the coming many-core era, improving programmability and
reliability of the system will be at least as important as
improving raw performance. To meet this vision, I will present a
processor feature called BugNet that assists programmers in
understanding software failures.

Reproducing software failures is a significant challenge. The
problem is severe especially for multi-threaded programs because
the causes of failure can be non-deterministic in nature. BugNet
continuously logs a program's execution while sacrificing very
little performance (~1%). If the program crashes, the developer
can use the log to debug the failure by time traveling to a past
instant in failed program's execution time. I will also talk
about two software tools built using the BugNet logging
mechanism. One is a data race analysis tool, which is currently
used at Microsoft. Another is a tool that enabled Intel to
easily port their ASIM architectural simulator to Mac OSX, which
was never attempted before because of the effort involved.
Satish Narayanasamy is a Ph.D. candidate in Computer Science at
the University of California, San Diego. His research interests
include computer architecture and systems support for software
development and system reliability. He has received two IEEE Top
Picks awards in recognition of his industry relevant
contributions.