        |
 |
|
|||
|
|
Silvius Rus - Research
Why a representation for memory location sets?
The representation of memory location sets in compilers is crucial to
the success of various analysis and optimization techniques:
data dependence, parallelization, locality enhancement, checkpoint etc.
For instance, a data independence test on a loop can be formulated as:
the set of memory location that carry dependences across different
iterations is empty.
Is it a hard problem?
In order to produce accurate information, a symbolic analysis
cannot resort to approximation.
It thus has to be closed with respect to composition operators
commonly used by analyses of memory references:
set intersection, set union,
set difference, predication,
union over an index space and
translation across subprogram boundaries (for interprocedural analyses).
Closure with respect to these operations guarantees tolerance to
failure of all analysis techniques that can be described with them.
What has been done before?
Most previous representations focused solely on systems of linear inequations.
They represented memory location sets as diophantine points in N-dimensional
polytopes. While this covers most code patterns found in small kernels,
analysis of larger program contexts results in shapes that do
not fit this representation.
What do we do?
We propose a representation that is closed to the operations listed above,
and that is persistent between compile-time
and run-time.
You can click here for a detailed description of the USR representation, or read the papers listed below.
Silvius Rus, Lawrence Rauchwerger, Jay Hoeflinger, "Hybrid Analysis: Static & Dynamic Memory Reference Analysis," International Journal of Parallel Programming, 31(4):251-283, Aug 2003. Also, In Proc. ACM Int. Conf.
Supercomputing (ICS), pp. 274-284, New York City, Jun 2002. Also, Technical Report, TR02-002, Parasol Laboratory, Department of Computer Science, Texas A&M University, Jan 2002.
Journal(ppt, abstract) Proceedings(ps, pdf, ppt, abstract) Technical Report(ps, pdf, ppt, abstract)
Parasol Home | Research | People | General info | Seminars | Resources Parasol Lab, 301 Harvey R. Bright Bldg, 3112 TAMU, College Station, TX 77843-3112 Contact Webmaster Phone 979.458.0722 Fax 979.458.0718 
Department of Computer Science | Dwight Look College of Engineering | Texas A&M University Privacy statement: Computer Science Engineering TAMU |