Graduate

ISL Colloquium: Delay, memory, and messaging tradeoffs in a distributed service system

Topic: 
Delay, memory, and messaging tradeoffs in a distributed service system
Abstract / Description: 

We consider the classical supermarket model: jobs arrive as a Poisson process of rate of lambda N, with 0 < lambda < 1, and are to be routed to one of N identical servers with unit mean, exponentially distributed processing times. We review a variety of policies and architectures that have been considered in the literature, and which differ in terms of the direction and number of messages that are exchanged, and the memory that they employ; for example, the "power-of-d-choices" or pull-based policies. In order to compare policies of this kind, we focus on the resources (memory and messaging) that they use, and on whether the expected delay of a typical vanishes as N increases.
We show that if (i) the message rate increases superlinearly, or (ii) the memory size increases superlogarithmically, as a function of N, then there exists a policy that drives the delay to zero, and we outline an analysis using fluid models. On the other hand, if neither condition (i) or (ii) holds, then no policy within a broad class of symmetric policies can yield vanishing delay.

Date and Time: 
Thursday, November 9, 2017 - 4:15pm
Venue: 
Packard 101

SCIEN & EE292E seminar: Interactive 3D Digital Humans

Topic: 
Interactive 3D Digital Humans
Abstract / Description: 

This talk will cover recent methods for recording and displaying interactive life-sized digital humans using the ICT Light Stage, natural language interfaces, and automultiscopic 3D displays. We will then discuss the first full application of this technology to preserve the experience of in-person interactions with Holocaust survivors

More Information: http://gl.ict.usc.edu/Research/TimeOffsetConversations/


The SCIEN Colloquia are open to the public. The talks are also videotaped and posted the following week on talks.stanford.edu.

There will a reception following the presentation.

Date and Time: 
Wednesday, November 8, 2017 - 4:30pm
Venue: 
Packard 101

EE380 Computer Systems Colloquium: Enabling NLP, Machine Learning, and Few-Shot Learning using Associative Processing

Topic: 
Enabling NLP, Machine Learning, and Few-Shot Learning using Associative Processing
Abstract / Description: 

This presentation details a fully programmable, associative, content-based, compute in-memory architecture that changes the concept of computing from serial data processing--where data is moved back and forth between the processor and memory--to massive parallel data processing, compute, and search directly in-place.

This associative processing unit (APU) can be used in many machine learning applications, one-shot/few-shot learning, convolutional neural networks, recommender systems and data mining tasks such as prediction, classification, and clustering.

Additionally, the architecture is well-suited to processing large corpora and can be applied to Question Answering (QA) and various NLP tasks such as language translation. The architecture can embed long documents and compute in-place any type of memory network and answer complex questions in O(1).

Date and Time: 
Wednesday, November 8, 2017 - 4:30pm
Venue: 
Gates B03

A New Class of Memory & Storage Technology: 3D XPoint™ and Optane™ Technology Overview

Topic: 
A New Class of Memory & Storage Technology: 3D XPoint™ and Optane™ Technology Overview
Abstract / Description: 

3D XPoint™ memory, the first new memory to ship in volume in decades, features the combination of DRAM and NAND attributes of: high memory density (NAND-like capacity); high performance (closer to DRAM performance); and non-volatility. The arrival of 3D XPoint™ memory – offering the full promise of a storage class memory – has the potential to fundamentally change the memory-storage hierarchy at the hardware, system software, and application levels. This memory is currently available as an Intel® Optane™ SSD with access times as fast as the rest of the system; a departure from classical storage technologies. System changes to match the low latency of these SSDs are already advanced, and in many cases they enable the application to utilize all the Optane SSD's performance. The implications of this new memory on computing are significant, with applications taking advantage of this new technology as storage as the first to benefit. Applications such as key–value stores and real-time analytics can benefit immediately both in terms of faster runtimes and also access to larger data sets through application or OS-based paging. To best measure this high performance SSD we borrow memory performance measurement techniques and apply them to storage. The next step in this convergence of memory and storage will be 3D XPoint memory accessed through processor load/store operations on DIMM busses.

Date and Time: 
Tuesday, November 7, 2017 - 4:00pm
Venue: 
Gates 415

SystemX Seminar: Reconfigurable platforms, the thirst for bandwidth, and the future of computing

Topic: 
Reconfigurable platforms, the thirst for bandwidth, and the future of computing
Abstract / Description: 

Today, designers of specialized systems are increasingly tantalized by the enormous energy efficiency of custom silicon solutions...but are just as turned off by the spiraling costs of building and verifying those very chips. Enter the FPGA, a reconfigurable substrate that would be the absolutely perfect solution were it not for two common conventional wisdoms: they're impossible to build, and impossible to use. In this talk we will discuss the motivation and context of FPGAs, notably from the perspective of IO circuits, and what interesting problems and solutions they pose to designers and users.

Date and Time: 
Thursday, November 9, 2017 - 4:30pm
Venue: 
Huang 018

Special Seminar: Topological Energy Transduction

Topic: 
Topological Energy Transduction
Abstract / Description: 

Within the CMOS architecture, the interconnected devices may either be categorized as an "active" device, which produces energy in the form of a current or a voltage, or a "passive" device, which stores or maintains energy in the form of a current or voltage. The societal demand for smaller sized electronic devices, such as computers and cellular phones, with improved functionality has forced not only the sizes of the constituent components of CMOS information processing technology to rapidly shrink, but for the operational frequencies to increase. While it has been possible to reduce the size of active CMOS devices, passive devices have not seen the same reduction in size. Of the passive devices (e.g. resistors, capacitors and inductors) used in CMOS technologies, the circuit element that consumes the most area on a circuit board while simultaneously finding the least success in miniaturization is the inductor. In this talk, we will present a novel method for energy transduction that utilizes the interplay between magnetism and topology on the surface of newly discovered materials, referred to as time-reversal invariant topological insulators, to create a paradigmatically different inductor. Using a novel self-consistent simulation that couples AC non-equilibrium Green functions to fully electrodynamic solutions of Maxwell's equations, we demonstrate excellent inductance densities up to terahertz frequencies thereby providing a potential solution to an eminent grand challenge.

Date and Time: 
Friday, November 3, 2017 - 12:00pm
Venue: 
Allen 338X

SmartGrid Seminar welcomes Saurabh Amin

Topic: 
TBA
Abstract / Description: 

The seminars are scheduled for 1:30 pm on the dates listed above. The speakers are renowned scholars or industry experts in power and energy systems. We believe they will bring novel insights and fruitful discussions
to Stanford. This seminar is offered as a 1 unit seminar course, CEE 272T/EE292T. Interested students can take this seminar course for credit by completing a project based on the topics presented in this course.


Yours sincerely,

Smart Grid Seminar Organization Team,

Ram Rajagopal, Assistant Professor, Civil & Environmental Engineering, and Electrical Engineering
Chin-Woo Tan, Director, Stanford Smart Grid Lab
Yuting Ji, Postdoctoral Scholar, Civil and Environmental Engineering
Emre Kara, Associate Staff Scientist, SLAC National Accelerator Laboratory

Date and Time: 
Thursday, November 16, 2017 - 1:30pm
Venue: 
Y2E2 111

SmartGrid Seminar: Optimization, Inference and Learning for District-Energy Systems

Topic: 
Optimization, Inference and Learning for District-Energy Systems
Abstract / Description: 

We discuss how Optimization, Inference and Learning (OIL) methodology is expected to re-shape future demand-response technologies acting across interdependent energy, i.e. power, natural gas andheating/cooling, infrastructures at the district/metropolitan/distribution level. We describe hierarchy ofdeterministic and stochastic planning and operational problems emerging in the context of physical flows over networks associated with the laws of electricity, gas-, fluid- and heat-mechanics. We proceed to illustratedevelopment and challenges of the physics-informed OIL methodology on examples of: a) Graphical Models approach applied to a broad spectrum of the energy flow problems, including online reconstruction of the grid(s) topology from measurements; b) Direct and inverse dynamical problems for timely delivery of services in the district heating/cooling systems; c) Ensemble Control of the phase-space cycling energy loads via Markov Decision Process (MDP) and related reinforcement learning approaches.

Date and Time: 
Thursday, November 2, 2017 - 1:30pm
Venue: 
Y2E2 111

SmartGrid Seminar: Emerging Technologies and Their Impact on the Grid

Topic: 
Emerging Technologies and Their Impact on the Grid
Abstract / Description: 

As rooftop solar, electric vehicles, and residential battery storage continue to become more and more commonplace, they can have significant impacts in the way the Energy Grid operates. By embracing these new technologies, PG&E is helping to create a vision for what a next generation energy company will look like and seeking to answer key questions such as: Is energy storage changing the way in which utilities operate the grid? What is needed for new technologies, such as residential battery energy storage, to go mainstream? What are some of the key factors driving the inevitable transition from a one-way grid to a two-way grid?

This presentation will focus on both the technology changes happening to the energy space as well as some of the technology advancements helping to reshape how the energy grid engages with these changes. It will cover these topics while exploring a case study of a recent pilot projects where PG&E, Tesla, GE & Green Charge teamed up on a project in San Jose to demonstrate how battery storage and rooftop solar connected to smart inverters can be used to support the electric grid during periods of high demand while providing participating residents and businesses with backup power and bill reduction. The project is a microcosm of what the grid will look like in the near future with the rapid adoption of distributed energy resources such as solar, battery storage & EVs.


 

The seminars are scheduled for 1:30 pm on the dates listed above. The speakers are renowned scholars or industry experts in power and energy systems. We believe they will bring novel insights and fruitful discussions
to Stanford. This seminar is offered as a 1 unit seminar course, CEE 272T/EE292T. Interested students can take this seminar course for credit by completing a project based on the topics presented in this course.

 

Yours sincerely,


Smart Grid Seminar Organization Team,

Ram Rajagopal, Assistant Professor, Civil & Environmental Engineering, and Electrical Engineering
Chin-Woo Tan, Director, Stanford Smart Grid Lab
Yuting Ji, Postdoctoral Scholar, Civil and Environmental Engineering
Emre Kara, Associate Staff Scientist, SLAC National Accelerator Laboratory

Date and Time: 
Thursday, November 9, 2017 - 1:30pm
Venue: 
Y2E2 111

Pages

Subscribe to RSS - Graduate