We describe our high-performance special purpose hardware system, Cube-4, for displaying and manipulating high resolution volumetric data sets in real-time. Cube-4 is designed to render 512x512x512, 16bit per voxel, data sets at 30 frames per second. Cube-4 implements a ray-casting algorithm in a highly-parallel and pipelined architecture. We also describe the recent Cube4 derivative Cube-4Light, which simplified the design but also added hardware support for perspective projections.
Implementation and Evaluation of Parallel Mesa Library
Tulika Mitra
In this talk, we describe the implementation and performance
evaluation
of a 3D graphics library that can be readily linked with
parallel applications to provide run-time visualization
on large-scale message-passing parallel machines, such as
Intel Paragon. The prototype implementation is currently fully
operational,
and is based on a public-domain OpenGL implementation Mesa,
and a sort-last parallelization strategy.
Through a detailed performance analysis, we show that the scalability
of the current prototype is close to the theoretical limit for
the given hardware architecture, and that further performance
improvement
on the same platform is unlikely to change the scalability characteristics,
especially when the number of rendering processors involved
is large.
We also develop a unified framework to describe
parallel compositing algorithms and show that two popular parallel
compositing
algorithms, binary swapping and parallel pipeline compositing,
are just two extreme instances of this framework. Such a framework
will
allow users to tailor the compositing algorithm according to
the
computation/communication characteristics of specific parallel
machines by tuning the
parameters appropriately. The current parallel Mesa library
prototype implements
such a parameterizable family of compositing algorithms.
Physically-based Animation of Volumetric Objects
Yan Chen
This paper presents a voxel-based animation technique which employs
either a mass-spring model or a finite element model.
Two volumetric
objects, a voxelized chair and a scanned muscle, are used as
case
studies with the different models. The mass-spring model
is used to
show an animation sequence of a falling and bouncing chair.
Wireframe
display and volume rendering are used to display a real-time
animation
of the process. In addition, a real-time simulation is
carried out by
the Finite Element Method (FEM) of a voxel-based multi-resolution
muscle mesh. Two techniques, a direct integration and
a simplified
modal analysis method are discussed in the context of applying
FEM for
muscle deformation. Local deformation optimization with
modal
analysis for higher resolution muscle volumetric animation,
which
allows accurate prediction of muscle deformation changes, has
been
used. Physiological muscle force has been considered and
a
biomechanically-based 3D FEM muscle model has been implemented.
Realistic animations have been produced based on the FEM simulation
with various graphics techniques.
Real Time Performance Guarantees over Wired/Wireless
LANs
Prashant Pradhan
Distributed audio and video applications require end-to-end
performance guarantees from the underlying networks for smooth
delivery of real-time media streams. The end-to-end nature
dictates
that real-time support be built into both wide-area and local-area
networks. With the enormous growth of portable computers equipped
with
networking capabilities, it is important to develop real-time
protocols for wireless LAN technologies, such that real-time
audio/video communication services can be extended to these
mobile
terminals as well. This paper describes the implementation effort
of
supporting end-to-end performance guarantees over a heterogeneous
local-area network comprising of both wired and wireless LAN
segments.
This work also addresses the performance guarantee issues associated
with terminal mobility in the wireless segments.
Development of Analytical and Simulation models for a Reliable Multicast Transport Protocol
Gursel Taskale
Model checking the i-protocol
Yifei Dong
Model checking is a verification technique aimed at determining
whether a system specification possesses a property expressed as
a temporal logic formula. The i-protocol is a bidirectional
sliding-window protocol employed in the GNU uucp file transfer utility.
I'll report how we verified the i-protocol and found a non-trivial livelock
error in it, as an example to introduce the conception, success, and challenge
of model checking.
Correct Execution at Different isolation Level
Shiyong Lu
The ANSI-92 specifications defines four levels of isolation:
READ UNCOMMITTED, READ COMMITTED, REPEATABLE READ and SERIALIZABLE.
One additional important isolation level, SNAPSHOT isolation,
is
also very popular in nowaday's commercial dababase managers.
In general, the weaker the isolation level, the greater the
degree of
concurrency allowed. However, only the SERIALIZABLE level provides
correct
behavior for all possible applications.
We believe that, for performance reasons, designers will continue
to execute
applications at isolation levels lower than serializable. Our
study of
such non-serializable schedules addresses two issues:
1. We would like to develop a technique for determining,
for a given
application, the least restrictive isolation
level consistent with
its semantics correctness. We are also
interested in the more
challenging problem of allowing different
transactions in the
same application to run at different
isolation levels.
2. In situations in which the semantics correctness of
an application
is not supported at a particular isolation
level, we want to
investigate techniques that can be used
within the transactions
themselves to provide additional synchronization
necessary to achieve
semantics correctness at that level.
In this talk, we present part of our research results for the
first issue.
Specifically, we give semantics conditions for different isolation
levels
such that, if a set of transactions satisfy the semantics condition
for
a particular isolation level, then we can run the set of transactions
at
that isolation level while semantics correctness is guaranteed.
Formal
proofs for these theorems are available upon request.
WebFlow Logic - A Logic for Everyday Workflow Practitioner
Hasan Davulcu
A workflow can simply be defined as a set of tasks that cooperate to
implement
a business process. Executing a workflow thus involves coordinated
execution
of multiple long-running activities in an environment of distributed,
heterogeneous processing entities.
We propose Concurrent Transaction Logic (CTR) as a declarative
language for
specifying, analyzing, and scheduling of workflows.
At present, we have a prototypical web-enabled system, WebFlow
Logic (WFL),
implemented in XSB (an efficient deductive database system).
The WFL system is tolerant to crashes and the workflow instances
in progress
are forward recoverable. CA's object-oriented database, Jasmine
is used as
the workflow database for storing and monitoring the progress
of active
workflow instances.
Tabled Constraint Logic Programming over Finite Domains
Baoqiu Cui
Constraint Logic Programming (CLP) is a new class of programming
languages which extend the declarativity of logic programming
by the
addition of constraint solving. In the past decade, CLP
has been
studied deeply in different directions and on several computation
domains. Just like in Logic Programming, tabling plays
an important
role in CLP. A CLP system with tabling techniques can
not only avoid
redundant computations and many infinite loops, but also give
an easy
way to find the optimal solutions for some problems. This
talk will
discuss some issues in applying tabling techniques into CLP
over finite
domains. An meta-interpreter of a CLP language over finite
domains,
which is written based on the XSB system, will be introduced.
Lan Huang
The topic is a compressed domain text retrieval system on the
internet, which supports approximate string search,efficient DB update,
and also allows time/space tradeoff .This system can be extended to other
applications like video annotation retrieval,technical report search.
A Multimedia Communication System
Sergio Silva
The basic idea of the Multimedia Communication System is to extend the
functionality of Internet Relay Chat (IRC) systems to include
other types
of media, such as voice and sketch image, to the usual text
interface.
The goal is to create a virtual study room, to allow students,
TAs and
professors to interactively discuss class related material through
communications networks, without being physically present.
Zodiac: A history-based video authoring system
Chuan-Kai Yang
Easy-to-use audio/video authoring tools play a crucial role in moving
multimedia programs from research curiosity to main-stream applications.
Recently we implement an interactive
video authoring system called Zodiac, which employs an innovative
edit history abstraction to support several unique editing features
not found in existing commercial and research video editing
systems.
Zodiac provides users a conceptually clean and semantically
powerful branching history model of edit operations to organize
the
authoring process, and to navigate among the design alternatives.
In addition, by analyzing the edit history, Zodiac is able
to reliably detect a composed stream's shot and scene boundaries,
which facilitate interactive video browsing. Zodiac also features
a video object annotation capability that allows users to
associate annotations to moving objects in a video sequence.
The annotations themselves could be text, image, audio, or video.
In this talk I will briefly talk about these three features
in our
system: branching history, shot/scene detection, video object
annotation.
EtheReal: A Framework for Bandwidth Guarantees over Switched Ethernet Network
Srinidhi Varadarajan
With advancement of technology and availability of powerful desktop
computers, human computer interaction using more natural data
sources like
video and audio has become commonplace. These multimedia sources
require
bandwidth guarantees from the network on which they travel.
Specifically,
multimedia data need to get from the source to the destination
in real time,
or within a predictable period of time. This is in contrast
to conventional
computer network traffic, such as file transfers or email, which
have no
stringent timing constraints. However, the most common LAN technology
in use
today, viz. Ethernet, treats both realtime and non-real-time
data
sources equally, and thus does not match the networking requirements
of
distributed multimedia applications.
The goal of the EtheReal project is to design and implement a
scalable Fast
Ethernet Switch that provides real time bandwidth guarantees
in a way that
is transparent to the host machines on the network. The
switch isolates
real-time sources from both non-real-time sources and other
real time
sources. It also prioritizes real time sources based upon their
requirements
from the network. A distinguishing feature of the work
is the capability to
provide these guarantees without any change to the existing
operating
system software or hardware on the hosts connected to the network.
The main
contribution of this research is the development of a novel
real-time
connection setup mechanism that assumes only common TCP/IP protocol
support
and is thus portable across multiple OS platforms.
An operational EtheReal prototype has been built using an off-the-shelf
PC
and commodity Ethernet cards. The hosts on the network can run
different
operating systems such as Windows 95, Linux and FreeBSD.
We are currently
working on fault tolerance mechanisms to continue switch operation
across
network hardware failures. The other major area of future research
is in
supporting real-time multicast, which is the enabling technology
for
Internet push technologies.
Performance Evaluation of An RSVP-Capable Router
Anindya Neogi
RSVP is a bandwidth reservation protocol that allows distributed
real-time
applications such as video-conferencing software to make bandwidth
reservations over packet-switched networks. Coupled with real-time
scheduling mechanisms built into packet routers, the network
guarantees
to provide the reserved bandwidth throughout the lifetime of
the
applications. Although guaranteed services are of great value
to both end users and carrier providers, their performance cost,
due
to additional control and data processing overhead, can potentially
have a negative impact on the packet throughput and latency
of the RSVP-capable routers. The goal of this paper is to examine
the performance cost of RSVP based on measurements from an
industry-strength RSVP implementation on a commercial IP router.
The focus is on the detailed evaluation of the performance implications
of various architectural decisions in RSVP.
We found that RSVP's control messages do not incur significant
overhead in terms of processing delay and bandwidth consumption.
However, the performance overhead of real-time packet scheduling
is
noticeable in the presence of a large number of real-time
connections. In extreme cases, the performance guarantees of
existing real-time connections may not be kept, and some best-effort
packets are actually dropped, although the overall bandwidth
requirement from these connections is smaller than the available
link bandwidth.
Return to the GRC home
page
If you have problems with this page,
please send E-mail to: