Supervisor: Prof. C.L. WANG
We have a few PhD (or 4-year 直博生) positions open for self-motivated and
academically strong students this year. If
you are interested in one of the projects, please contact me at
firstname.lastname@example.org. Interview will be
arranged for qualified students.
- 漫谈面向大数据,云计算平台建设的新视角与新技术 by C.L. Wang (05/10/2014):
Speculative Java Computing on GPUs (9/2014 --)
GPUs open up new opportunities for accelerating the Java
programs for high-speed big data analytics. In this new
project, we will develop a portable Java library and runtime
environment "Japonica+" for supporting GPU acceleration of
auto-parallelized loops with non-deterministic data
dependencies. The runtime can support parallelization of a
sequential Java program (with non-deterministic data
dependencies) into parallel workloads (either Java threads or
OpenCL x86 kernels) to run on CPU and OpenCL kernels to run on
GPU concurrently, utilizing all the CPU and GPU computing
resources. Task to be done: (1) automatic translation
from Java bytecode to OpenCL, (2)
auto-parallelization of loops with non-deterministic data
dependencies (See our GPU-TLS paper), (3) dynamic load scheduling and
rebalancing via task migration between CPU and GPU,
(4) virtual shared memory support between host and
multi-GPU cards. The whole project will be developed in recent
Nvidia K40 GPU cards.
1-2 Ph.D students (first priority):
Solid backgroudnd in compiler techniques (e.g., loop
parallelization, dependency checking), GPU hardware
architecture (Nvidia or AMD GPUs), and good experiences in GPU
programming (CUDA or OpenCL).
can apply any time if you have the
above experiences (especially OpenCL)
- Nvidia Tesla K40 GPU :
- HKU GPU Cluster: 12 x IBM iDataPlex dx360 M3 server
connected by a QDR IB switch, each has one Nvidia M2050
GPU. (See :
- Previous Japonica
(Java with Auto-Parallelization ON GraphIcs
Co-processing Architecture) project:
- Guodong Han, Chenggang Zhang, King Tin Lam, and Cho-Li Wang,
Java with Auto-Parallelization on Graphics Coprocessing Architecture, 42nd International Conference on Parallel Processing (ICPP2013), October 1-4, 2013, Lyon, Lyon, France.
- Chenggang Zhang, Guodong Han, Cho-Li Wang,
An Efficient Runtime for Speculative Loop Parallelization on GPUs, 13th
IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid),
May 13--16, 2013, Delft, The Netherlands. (pdf)
- Previous projects on Distributed Java Virtual Machines:
TrC-MC: Software Transactional memory on multi-core.
Runtime Support with Software Coherence for Future 1000-Core
Tiled Architectures, HKU 716712E,
9/2012-8/2015, supported by HK RGC.
Moving up to a parallelism with
1,000 cores requires a fairly radical rethinking of how to
design system software. With a growing number of cores,
providing hardware-level cache coherence gets increasingly
complicated and costly, leading researchers to promote
abandoning it if future many-core architectures are to stay
inherently scalable. That means software now has to take on
the role in ensuring data coherence among cores. In this research,
we address the above issues and propose novel methodologies to
build a scalable CoC ("Cloud on Chip") runtime platform, dubbed
Runtime with Object Coherence On Dynamic tILES), for
future 1000-core tiled processors.
involves the development of two important software
(1) Cache coherency protocol (2)
DVFS-based power management. (Currently, 3 Ph.D students
are working on this project.)
student: strong background in OS kernel, full knowledge
in memory subsystem (cache/DRAM, paging), cache coherent
Require strong background in
software distributed shared memory systems (e.g.,
programming experiences in multicore power management systems.
“Rhymes: A Shared Virtual
Memory System for Non-Coherent Tiled Many-Core Architectures,”
to appear in ICPADS2014.
- “Latency-aware Dynamic
Voltage and Frequency Scaling on Many-core Architecture for
CloudCom-Asia 2013, Fuzhou, China, Dec. 16-18, 2013.
and In-Memory Computing (内存计算)
- Performance Optimization of Apache Spark on
SDN-connected Cluster (Now: 2 M.Sc students + one PhD
- One M.Sc student (Mr. Ying Li) is now implementing
"Nesox", a network resource scheduler for data-parallel
computing by leveraging SDN techniques.
1 Ph.D student:
must have strong interest in Cloud, Network Virtualization
Operating System for Manycore Systems
(“NoHype” Cloud Operating System
Traditional operating systems are based on the
sequential execution model developed in the 1960s. Such
operating systems cannot address new many-core parallel
hardware architecture without major redevelopment. For
instance, how can you harness the power a next-generation
manycore processor with >1,000 cores? We will investigate
various perspectives on the future OS design towards the goal.
We are developping an x86-based full-system simulator based on
Gem5. (Currently, one Ph.D student is working
on this project.)
1 Ph.D student: must have some experiences in OS kernel development.
Some experiences in Barrelfish or sccLinux will be quite
helpful. The student is also required to have good knowledge in multicore
Updated: Sept. 06, 2014