Hangchen Yu, Arthur M. Peters, Amogh Akshintala, Christopher J. Rossbach: AvA: Accelerated Virtualization of Accelerators, ASPLOS’20, Mar 2020. [PDF][POSTER(IAP’19 Best Poster Award)]
AvA provides automated construction of support for hypervisor-mediated accelerator sharing among mutually distrustful VMs. AvA combines a DSL for describing accelerator APIs and sharing policies, a device-agnostic runtime, and tools to generate and deploy accelerator-specific stack components such as guest libraries and API servers. AvA uses a novel technique called Hypervisor Interposed Remote Acceleration (HIRA) that retains hypervisor interposition for efficient policy enforcement.
Amogh Akshintala*, Hangchen Yu*, Arthur M. Peters, Christopher J. Rossbach: Trillium: The code is the IR, VIRT’19, July 2019. [PDF]
We hypothesize that while API-remoting may be the only viable software virtualization technique, it should not be implemented purely in user-space. We find that hypervisor-mediated API-remoting can be efficient: Decoupling device virtualization from GPU ISA virtualization is key to preserving the raw speedup from GPGPU acceleration.
Hangchen Yu, Arthur M. Peters, Amogh Akshintala, Christopher J. Rossbach: Automatic Virtualization of Accelerators, HotOS’19, May 2019. [PDF][POSTER(OSDI’18)][DOI]
AvA (Automatic Virtualization of Accelerators) re-purposes a para-virtual I/O stack design based on API remoting to present virtual accelerator APIs to guest VMs. With AvA, a single developer could virtualize a core subset of OpenCL at near-native performance in just a few days.
Youngjin Kwon, Hangchen Yu, Simon Peter, Christopher J. Rossbach, Emmett Witchel: Ingens: Huge Page Support for the OS and Hypervisor, SIGOPS OSR’17, August 2017. [PDF][DOI]
Ingens is a framework for providing transparent huge page support in a coordinated way. Ingens manages contiguity as a first-class resource, and tracks utilization and access frequency of memory pages, enabling it to eliminate pathologies that plague current systems.
Hangchen Yu, Christopher J. Rossbach: Full Virtualization for GPUs Reconsidered, WDDD’17, July 2017. [PDF]
This paper revisits and remeasures GPUvm, a Xen-hypervisor-based full virtualization design for supporting VM access to discrete NVIDIA GPUs. While we are able to reproduce some reported results, we also reach a number of contrary findings. For example, full virtualization introduces catastrophic overhead in initialization and GPUvm’s optimizations; the BAND scheduler algorithm increases maximum unfairness by up to 6%, at the cost of decreasing aggregate throughput by as much as 8%.
Youngjin Kwon, Hangchen Yu, Simon Peter, Christopher J. Rossbach, Emmett Witchel: Coordinated and Efficient Huge Page Management with Ingens, OSDI’16, Nov. 2016. [PDF][POSTER]
Ingens is a framework for huge page support that relies on a handful of basic primitives to provide transparent huge page support in a principled, coordinated way. By managing contiguity as a first-class resource and by tracking utilization and access frequency of memory pages, Ingens is able to eliminate a number of fairness and performance pathologies that plague current systems.
Chong He, Hangchen Yu, Xianling Liang, Junping Geng, Ronghong Jin: Sideband Radiation Level Suppression in Time-modulated Array by Nonuniform Period Modulation: AWPL’14, Nov. 2014. [PDF][DOI]
A novel sideband radiation (SR) level suppression method for the time-modulated array (TMA) is proposed. By modulating their front-end RF switches with nonuniform periods, the SR level of the TMA can be remarkably suppressed, compared to using uniform periods.
Liang Liu, Ronghong Jin, Hangchen Yu, Xianling Liang, Junping Geng, Xudong Bai: A Compact Ultra-wideband Power Divider with High Isolation, APSURSI’14, July 2014 [PDF][DOI]
A novel ultra-wideband divider with high isolation is proposed, which bases on the multi-section transmission line theory to extend the bandwidth of the power divider and the multilayer technique to reduce the size of the power divider.
Yaxiong Liu, Cailian Chen, Hangchen Yu, Xinping Guan: Distortion Analysis for Delay Tolerant Data Collection for High-speed Wireless Sensor and Actor Networks, WCICA’12, July 2012. [PDF][DOI]
In this paper, we propose to leverage high-speed rail as mobile actor to assist structural health monitoring (SHM) data collection and delivery for wireless sensor and actor networks (WSANs).
>> Read more: [Google Scholar]
VIRT’19 Trillium: The code is the IR, Dublin, Ireland, Jul 18, 2019 [SLIDES]
We propose the unnecessity and negative effects of the virtual ISA in the virtualization stack.
HotOS’19: Automatic Virtualization of Accelerators, Bertinoro, Italy, May 10, 2019. [SLIDES] (Presenter: Arthur M. Peters)
We demonstrate the feasibility of automatic construction of virtual accelerator stacks.
WDDD’17: Full Virtualization for GPUs Reconsidered, Toronto, Canada, Jun 25, 2017. [SLIDES]
This talk revisits the GPU full-virtualization technique using GPUvm as the example system.
The publication list contains most of the research projects, while there are a handful of sophisticated course projects worth being mentioned here.
Due to the heavy schedule of projects, I plan to complete this section at some point in the future. At this moment, please reference CV.pdf for more details.
[NOT YET ADDED]