| Jay Sipelstein. Data representation optimizations for collection-oriented languages. PhD thesis, School of Computer Science, Carnegie Mellon University, to appear. |
....compiler. However, the compiler can reduce some of these effects through optimizations that trade off load imbalance vs. redundant computation. One such optimization is the use of a lazy pack that delays the actual packing of a vector until the load imbalance rises above a tolerance. Sipelstein [61] is studying such optimizations. 6 Performance Evaluation This section presents performance numbers for several benchmarks coded in VCODE, translated to cthreads code using the compiler, and executed on an Encore Multimax running the Mach operating system. The benchmark programs emphasize ....
SIPELSTEIN,J.M.Data Representation Optimizations for Collection-Oriented Languages. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA, 1992. In preparation.
....among the processors to properly maintain the stack. How to keep this ordering and how much the heuristic really helps is an area of current research. Another optimization would be to reschedule computation threads only when there is a significant load imbalance among the processors (see, e.g. Sip95] An open question is whether our algorithm for time , work , and space efficient execution of nested parallel programs can be extended so as to guarantee provably communication efficient execution as well. ....
J. Sipelstein. Data Representation Optimizations for Collection-Oriented Languages. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA, 1995. To appear.
No context found.
Sipelstein, J. Data Representation Optimizations for Collection-Oriented Languages.PhD thesis, School of Computer Science, Carnegie Mellon University. To appear. 27
....This is important for guaranteeing termination. The communication costs involved in doing the packing and unpacking can be quite high, and one area of our ongoing research is to see how the communication can be reduced by only packing when there is a significant load imbalance among processors [53]. We now briefly describe the third phase of the NESL compiler, which involves specializing polymorphic code. The standard implementation of polymorphism is to have a single compiled version of a function and achieve polymorphism by use of pointers the function only manipulates pointers to the ....
Jay Sipelstein. Data Representation Optimizations for Collection-Oriented Languages. PhD thesis, School of Computer Science, Carnegie Mellon University. To appear.
....This is important for guaranteeing termination. The communication costs involved in doing the packing and unpacking can be quite high, and one area of our ongoing research is to see how the communication can be minimized by only packing when there is a significant load imbalance among processors [38]. CVL: To enable us to quickly implement Vcode on new machines, we designed Cvl (C Vector Library) a library of low level segmented vector routines callable from C. These are used by a Vcode interpreter, described below. Our implementations of Cvl on the Connection Machine CM 2, Cray and serial ....
J. Sipelstein. Data Representation Optimizations for Collection-Oriented Languages. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA, To appear.
No context found.
Jay Sipelstein. Data representation optimizations for collection-oriented languages. PhD thesis, School of Computer Science, Carnegie Mellon University, to appear.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC