Brian Demsky Associate Professor Electrical Engineering and Computer Science University of California, Irvine email: office: EH 4414 phone: (949) 824-0356 fax: (949) 824-3203

Research Interests

• Bamboo: Software Design for Multi- and Many- Core Processors: Developing software for future multi-core processors will require new software design methodologies, programming languages, and tools. This research investigates a new design methodology for developing the parallel software systems that are necessary to take advantage of multi-core processors. This new design process enables the software developer to create flexible system designs that easily accommodate refinement of how the computation is realized. It does this by separating the functional design of the software system from the specification of how to organize the computation. This research project investigates the construction of synthesis and profiling tools that can be used to develop and refine these functional and organizational specifications. These specifications are in turn used to create an executable that is optimized for the specific multi-core microprocessor.
  
  
• Disjointness Analysis: We have developed a new disjointness analysis for Java-like languages. Two objects are disjoint if the parts of the heap reachable from the two objects are disjoint. The analysis can be used verify program correctness properties and to enable a new approach to automatically parallelizing applications.
  
  
• OoOJava: We have developed a new approach to automatic parallelization that combines a software-based implementation of superscalar scheduling with disjointness analysis.
  
  
• Transactional Distributed Shared Memory: We have developed a new transaction-based approach to distirbuted shared memory. Our transaction based approach makes aggressive use of both prefetching and caching of remote objects to hide network latency while relying on the transaction commit mechanism to preserve the simply transactional consistency model.
  
  
• I/O for Transactional Memory: We have developed an I/O implementation for software transactional memory systems. Our implementation allows transactions to safely make use of file operation inside of memory transactions.
  
  
• Bristlecone: Robust Software Systems: We have a developed a language-based approach for building robust software systems. This approach decomposes software systems into a set of a tasks, and uses task specifications to describe how to combine these tasks. In the event of a failure, our system reasons about these task specifications to adapt the execution and continue to provide service.
  
  
• Garm: Protecting Private Data in Legacy Applications: We have developed Garm, a new tool for tracing data provenance and enforcing data access policies with arbitrary binaries. Users can use Garm to attach access policies to arbitrary data files (or parts of files) and Garm ensures that all accesses to the data across all applications and executions are consistent with the policy.
  
  
• Analysis of Software Bugs: We have built a prototype tool to automatically build a repository of software bugs. For each software bug, our tool records a test case that reveals the bug, a version of the program that contains the bug, and the change that corrects the software bug. We are in the process of using this tool to collect information and plan to mine this information to discover properties of software bugs.
  
  
• Data Structure Repair: We have developed data structure repair, a technique to enable programs to recover from data structure corruption errors. This technique can be applied to both volatile data structures inside a running program and persistent data structures stored on disk.

Software

Courses

Select Publications

Presentations

Other