Yang Hu - 1

CURRICULUM VITAE - YANG HU

CONTACT INFORMATION

Ph.D. candidate

Phone: (352) 281-3364

huyang.ece@ufl.edu

URL: http://plaza.ufl.edu/huyang.ece

#327 Benton Hall

University of Florida

Gainesville, FL 32611

EDUCATION

Ph.D. University of Florida Aug. 2011 – Aug. 2017 (expected)

Ph.D. Student in Computer Engineering

M.S. Tsinghua University, China Sep. 2008 – Jun. 2011

Graduate Student in Microelectronic Engineering

B.S. Tianjin University, China Sep. 2003 – Jul. 2007

Undergraduate Student in Electronic Engineering

AWARDS AND HONORS

Best Paper Nomination of HPCA 2017 2017

Best of IEEE Computer Architecture Letters Award 2015

UF Alumni Fellowship 2011~2015

Travel Grant for 49th Int'l Symp. on Micro Architecture (MICRO) 2016

Travel Grant for 42th Int'l Symp. on Computer Architecture (ISCA) 2015

Travel Grant for 46th Int'l Symp. on Micro Architecture (MICRO) 2013

Travel Grant for 39th Int'l Symp. on Computer Architecture (ISCA) 2012

ACADEMIC TALKS

“Towards Efficient Server Architecture for Virtualized Network Function Deployment” MICRO, 2016

“HOPE: Enabling Efficient Service Orchestration in Software-Defined Data Centers” ICS, 2016

“Towards Sustainable in-Situ Server Systems in the Big Data Era” ISCA, 2015

PROFESSIONAL SERVICES

Program Committee Member: The 4th Workshop on Efficient Data Center Systems (EDCS), 2016

Journal Reviewer: IEEE Transactions on Parallel and Distributed Systems (TPDS)

IEEE Computer Architecture Letters (CAL)

Conference Reviewer: ISCA, HPCA, MICRO, ASPLOS, ICS, ICDCS, ICPP

Workshop Reviewer: ASBD, CoolDC

SELECTED PUBLICATIONS

1. Yang Hu, Mingcong Song, and Tao Li. “Towards Efficient Service Chaining in Heterogeneous Network

Function Virtualization Deployment”. To submit, ISCA 2017

2. Yang Hu and Tao Li. “Towards Architectural Support for “Full Containerization” in Container Based Network

Function Virtualization”. To appear, ASPLOS 2017. (Acceptance Rate: 53/306 = 17%)

3. Mingcong Song, Yang Hu, Huixiang Chen, and Tao Li. “Towards Pervasive and User Satisfactory CNN across

GPU Microarchitectures”. To appear, HPCA 2017. (Best Paper Nomination, ranked No.1 in 231 submissions)

Yang Hu - 2

4. Chao Li, Yang Hu, Juncheng Gu, Jingling Yuan, and Tao Li. “Oasis: Scaling Out Datacenter Sustainably and

Economically”. Accepted, Transactions on Parallel and Distributed Systems (TPDS), 2016, IEEE

5. Yang Hu, Mingcong Song, and Tao Li. “Towards Efficient Server Architecture for Virtualized Network

Function Deployment: Implications and Implementations”. Proc. of the 49th International Symposium on

Microarchitecture (MICRO), 2016. (Acceptance Rate: 61/288 = 21.2%)

6. Mingcong Song*, Yang Hu*, and Tao Li. “Bridging the Semantic Gaps of GPU Acceleration for Scale-out

CNN-based Big Data Processing: Think Big, See Small”. Proc. of the 25th International Conference on Parallel

Architectures and Compilation Techniques (PACT), 2016 (Authors with equal contribution, acceptance Rate:

31/139 = 22.3%)

7. Yang Hu, Chao Li, Longjun Liu, and Tao Li. “HOPE: Enabling Efficient Service Orchestration in Software-

Defined Data Centers”. Proc. the 30th International Conference on Supercomputing (ICS), Jun. 2016

(Acceptance Rate: 43/178 = 24.2%)

8. Longjun Liu, Chao Li, Hongbin Sun, Yang Hu, Tao Li, Jingmin Xin and Nanning Zheng.. “Towards an

Adaptive Multi-Power-Source Datacenter”. Proc. the 30th International Conference on Supercomputing (ICS),

Jun. 2016 (Acceptance Rate: 43/178 = 24.2%)

9. Chao Li*, Yang Hu*, Longjun Liu*, Juncheng Gu, Mingcong Song, Jingling Yuan, Xiaoyao Liang, and Tao

Li. “Towards Sustainable in-Situ Server Systems in the Big Data Era”. Proc. the 42nd ACM/IEEE International

Symposium on Computer Architecture (ISCA), Jun. 2015 (Authors with equal contribution) (Acceptance Rate:

58/305 = 19%)

10. Longjun Liu*, Chao Li*, Hongbin Sun, Yang Hu, Juncheng Gu, Tao Li, Jingmin Xin and Nanning Zheng.

“HEB: Deploying and Managing Hybrid Energy Buffers for Improving Datacenter Efficiency and Economy”.

Proc. the 42nd ACM/IEEE International Symposium on Computer Architecture (ISCA), Jun. 2015 (Authors

with equal contribution) (Acceptance Rate: 58/305 = 19%)

11. Longjun Liu, Chao Li, Hongbin Sun, Yang Hu, Juncheng Gu, and Tao Li. “BAAT: Towards Dynamically

Managing Battery Aging in Green Datacenters”. Proceedings of the 45th Annual IEEE/IFIP International

Conference on Dependable Systems and Networks (DSN), Jun. 2015 (Acceptance Rate: 50/227 = 22%)

12. Longjun Liu, Chao Li, Hongbin Sun, Yang Hu, Jingmin Xin, Nanning Zheng, and Tao Li. "Leveraging

Heterogeneous Power for Improving Datacenter Efficiency and Resiliency". IEEE Computer Architecture

Letters (CAL), 2015 (Best of Computer Architecture Letters)

13. Chao Li, Rui Wang, Yang Hu, Ruijin Zhou, Ming Liu, Longjun Liu, Jingling Yuan, Tao Li, and Depei Qian,

“Towards Automated Provisioning and Emergency Handling in Renewable Energy Powered Datacenters”,

Journal of Computer Science and Technology (JCST), Vol. 29, No. 4, pp. 618-630, 2014. Springer

14. Chao Li, Yang Hu, Ruijin Zhou, Ming Liu, Longjun Liu, “Enabling Datacenter Servers to Scale Out

Economically and Sustainably”. Proc. of the 46th Intl. Symposium on Microarchitecture MICRO 2013.

(Acceptance Rate: 39/269 = 14.5%)

15. Shouyi Yin, Yang Hu, Zhen Zhang, Shaojun Wei, “Hybrid Wired/Wireless On-Chip Network Design for

Application-Specific SoC”, IEICE transactions on electronics 95 (4), 495-505, 2012

16. Yang Hu, Shouyi Yin, Shaojun Wei, “Mixed-level modeling for network on chip infrastructure in SoC

design”, IEEE Asia Pacific Conference on Circuits and Systems (APCCAS), 2010

Yang Hu - 3

PATENTS

Chinese Patent, 201510553330, Filed, 2015

Chinese Patent, 201510556261, Filed, 2015

Chinese Patent, 201510558202, Filed, 2015

SELECTED RESEARCH PROJECTS AND EXPERIENCE

Optimizing the Networking Performance on Containerized NFV Platform Dec. 2015~Now Goal: Explore the micro-architectural bottlenecks in Linux container based NFV platform. Improving the

performance of DPDK based containerized NFV platform.

Optimizing the Virtual Switch Performance on NUMA based NFV Platform Oct. 2014~Dec. 2015 Goal: Explore the performance bottlenecks of NUMA based COTS server in modern NFV deployment.

Optimize the thread scheduling of Open vSwitch.

o Characterize the performance impacts of NFV data path critical threads (ksoftirq, vhost-net) mapping

on NUMA based COTS server. Develop the architectural performance slowdown model for NFV data

path using hardware performance counter readings.

o Optimize the thread scheduling scheme of Open vSwitch to minimize the aggregated slowdown of

NFV traffic flows.

Orchestrate the Management Workloads in Software-Defined Data Center Jun. 2014~Oct. 2014 Goal: Design a cloud management system to coordinate the power consumption of management workloads in

modern software-defined data center.

o Characterize the power impacts of management workloads in a modern Software-Defined Data Center.

Explore the performance/power impacts of inter-machine management workloads to the server cluster.

o Develop a graph based algorithm to analyze and model the inter-machine power impacts of

management workloads.

o Implement a framework that integrated in OpenStack platform to mitigate the inter-machine power

impacts of SDDC management workloads.

Power/Performance Optimization in a Solar Powered Cloud Data Center Jan. 2014~May. 2014 Goal: Develop renewable energy penetrated cloud data center with power-aware cloud management.

o Build a full-fledged 8-node private cloud data center prototype based on OpenStack Havana. Nova,

Neutron, Cinder and Swift are deployed. Metering service (Telemetry) is also deployed. The data

center is powered by our roof-mounted PV panels and integrated inverter system.

o Implement a data collector in Python that collects power data from power delivery infrastructure

(utility and renewable energy) and IT equipment. It also communicates with other OpenStack services.

Monitoring tool (Zabbix) is deployed to collect performance data.

o Implement a cross-level power management framework and integrated it into OpenStack. The power

management framework coordinates the data center performance and solar power generation. A nova

scheduler is implemented in the power management framework.

Design In-Situ Data Processing Server System for Rural Area Sep. 2013~Dec. 2013

Yang Hu - 4

Goal: Implement an off-grid system that aims for the data processing at rural area. The system could save

considerable communication expenditure.

o Design a spatial-temporal power management scheme to balance the batteries’ lives in a battery array.

o Build a battery switch array using contactor and Siemens PLC. The central controller schedules the

battery array configuration through HMI.

Build a Grid-tie Solar Powered Data Center Feb. 2013 ~Aug. 2013 Goal: Build the experimental platform for lab’s green computing project.

o In charge of the whole purchase and construction of the project.

o Implement the data collector that communicates with Inverter, MPPT and HMI using Modbus TCP

protocol.

Implement Simulation Platform of Virtual Desktop Infrastructure Sep. 2011~Nov. 2012 Goal: Explore the architecture throttling when deploying VDI service.

o Build the simulation framework using Qemu and DiskSim 2.0.

o Running VDI framework based on RDS on the simulation platform.

Design Transaction-Level Simulator for Intel Advanced SoC Prototype Jan. 2009~May. 2011 Goal: Build a simulation platform for pre-silicon verification.

o Implement generic template of SoC peripherals using SystemC TLM 2.0.

o Implement the interconnection as NoC architecture.