Select a calendar:
Filter May Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for May 11, 2016
-
Understanding and Improving Graph Algorithm Performance
Wed, May 11, 2016 @ 10:30 AM - 11:30 AM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Mr. Scott Beamer, Computer Architecture PhD Candidate/UC Berkeley
Talk Title: Understanding and Improving Graph Algorithm Performance
Abstract: Graph processing is experiencing a renewed surge of interest as applications grow in importance in social networks analysis, recognition and the sciences. Unfortunately, graph algorithms often execute inefficiently on today's hardware platforms. In particular, graph algorithms are often able to simultaneously underutilize a platform's compute throughput and memory bandwidth.
In this talk, I will describe my work to identify these hardware bottlenecks and their causes. By understanding the main factors for graph algorithm performance, we can design hardware better suited for graph algorithms or improve graph algorithm software implementations. Through our characterization work, we find that contrary to the notion that graph algorithms have a random memory access pattern, we find well-tuned parallel graph codes exhibit substantial locality and thus experience a moderately high cache hit rate.
To understand these graph processing bottlenecks and the solutions to them requires a vertically integrated approach, ranging from algorithms with a novel breadth-first search algorithm to architecture with a detailed graph workload characterization. In between, this includes a graph domain-specific language, a performance model, and a benchmark suite. I will conclude the talk with our recent work on reducing memory communication via an algorithmic transformation.
Biography: Scott Beamer is a Computer Architecture PhD candidate at UC Berkeley advised by Krste Asanovic and David Patterson. He is currently investigating how to accelerate graph algorithms through software optimization and hardware specialization. In the past, he looked into how to best use monolithically integrated silicon photonics to create memory interconnects. He received a B.S. in Electrical Engineering and Computer Science and a M.S. in Computer Science, both from UC Berkeley.
Host: Professor Sandeep K. Gupta
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Mayumi Thrasher
-
PhD Defense - Andrew Jones - Rendering for Automultiscopic Displays
Wed, May 11, 2016 @ 03:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Andrew Jones, PhD Candidate
Talk Title: Rendering for Automultiscopic Displays
Abstract: Title: Rendering for Automultiscopic Displays
Location: SAL 322
Time: 3:00pm - 5:00pm, May 11th, 2016
PhD Candidate: Andrew Jones
Committee Members:
Paul Debevec
Mark Bolas (outside member)
Jernej Barbiq
Abstract:
While a great deal of computer generated imagery is modelled and rendered in three dimensions, the vast majority of this 3D imagery is shown on two-dimensional displays. Various forms of 3D displays have been contemplated and constructed for at least one hundred years, but only recent advances in digital capture, computation, and display have made functional and practical 3D displays possible. In this thesis, I propose several designs that overcome some of the classic limitations of 3D displays. The displays are: autostereoscopic, requiring no special viewing glasses; omnidirectional, allowing viewers to be situated anywhere around it; and multiview, producing a correct rendition of the 3D objects with correct horizontal parallax and vertical perspective for any viewer around the display.
The first display prototype utilizes a spinning anisotropic mirror to distribute frames from a high-speed video projector to different viewers. Unfortunately, as the size and mass of the mirror increases, it becomes increasingly difficult to maintain a stable and rapid rotation speed. The second 3D display form has no moving mechanical parts, provides interactive content, and scales to large format displays. The key insight is that a large array of closely stacked projectors aimed at a stationary anisotropic screen is optically equivalent to a single high-speed projector aimed at a rotating anisotropic screen. Both types of display utilize new algorithms based on geometry and light filed based rendering. Applications for these displays include life-size interactive virtual characters, 3D teleconferencing, and time-offset conversations with 3D subjects.
Host: Andrew Jones
Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Ryan Rozan
