2010 Poster Sessions : GRAMPS: A Programming Model for Heterogeneous Many-Core Parallelism

Student Name : Jeremy Sugerman
Advisor : Patrick Hanrahan
Research Areas: Graphics/HCI
Abstract:
GRAMPS is a programming model where parallel applications are expressed as pipelines and graphs of independent stages communicating via queues.
Stages can be task or data-parallel and target programmable or custom cores. This poster describes two GRAMPS studies: graphics pipelines on simulated GPU-like and CPU-like systems and a range of applications including ray tracing, MapReduce, and physical simulation, on a run-time for a real machine with 8 SMT Nehalem cores.

Bios:
Jeremy Sugerman is a PhD candidate in CS at Stanford.

David Lo is a coterminal student in EE at Stanford.

Daniel Sanchez is a PhD candidate in EE at Stanford.

Richard Yoo is a PhD candidate in EE at Stanford.