Hyoseung Kim

Hyoseung Kim

Auto-generated profile, based on publicly available information

Hyoseung Kim

University of California, Riverside

Research Focus & Metrics

Academic Career Length

21Years (2006-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 April 2014
Verified
Conference Paper
Closed access
HKHyoseung Kim
Hyoseung Kim
DNDionisio de Niz
Dionisio de Niz
BABjorn Andersson
Bjorn Andersson
MKMark Klein
Mark Klein
OMOnur Mutlu
Onur Mutlu
RRRagunathan Rajkumar
Ragunathan Rajkumar
In commercial-off-the-shelf (COTS) multi-core systems, a task running on one core can be delayed by other tasks running simultaneously on other cores due to interference in the shared DRAM main memory. Such memory interference delay can be large and highly variable, thereby posing a significant challenge for the design of predictable real-time systems. In this paper, we present techniques to provide a tight upper bound on the worst-case memory interference in a COTS-based multi-core system. We explicitly model the major resources in the DRAM system, including banks, buses and the memory controller. By considering their timing characteristics, we analyze the worst-case memory interference delay imposed on a task by other tasks running in parallel. To the best of our knowledge, this is the first work bounding the request re-ordering effect of COTS memory controllers. Our work also enables the quantification of the extent by which memory interference can be reduced by partitioning DRAM banks. We evaluate our approach on a commodity multi-core platform running Linux/RK. Experimental results show that our approach provides an upper bound very close to our measured worst-case interference.

1 December 2019
Verified
Conference Paper
Closed access
YXYecheng Xiang
Yecheng Xiang
HKHyoseung Kim
Hyoseung Kim
Deep neural networks (DNNs) have been showing significant success in various applications, such as autonomous driving, mobile devices, and Internet of Things. Although much research has been conducted to optimize the structure of DNNs, limited attention has been given to their timely execution, specifically on the scheduling of real-time inference requests to various DNN models. For instance, existing DNN frameworks, such as Caffe, TensorFlow and Torch, only provide a single-level priority, one-DNN-per-process execution model and sequential inference interfaces. They can be particularly problematic when used in edge computing and in-vehicle intelligence systems for multiple DNNs, as response time may become unpredictably long in the worst case while leaving system resources underutilized. This paper presents DART, a DNN scheduling framework that offers deterministic response time to real-time tasks and increased throughput to best-effort tasks. DART employs a pipeline-based scheduling architecture with data parallelism, where heterogeneous CPUs and GPUs are arranged into nodes with different parallelism levels. DART also includes pipeline stage design and node configuration schemes, admission control, execution time profiling, and runtime enforcement techniques. We evaluated DART on Intel x86 Xeon and Nvidia ARM platforms with GPUs. Experimental results indicate that DART significantly outperforms the existing approaches, by up to 98.5% shorter worst-case response time for real-time tasks while simultaneously achieving up to 17.9% higher throughput for best-effort tasks.

1 July 2013
Verified
Conference Paper
Closed access
HKHyoseung Kim
Hyoseung Kim
AKArvind Kandhalu
Arvind Kandhalu
RRRagunathan Rajkumar
Ragunathan Rajkumar
Many modern multi-core processors sport a large shared cache with the primary goal of enhancing the statistic performance of computing workloads. However, due to resulting cache interference among tasks, the uncontrolled use of such a shared cache can significantly hamper the predictability and analyzability of multi-core real-time systems. Software cache partitioning has been considered as an attractive approach to address this issue because it does not require any hardware support beyond that available on many modern processors. However, the state-of-the-art software cache partitioning techniques face two challenges: (1) the memory co-partitioning problem, which results in page swapping or waste of memory, and (2) the availability of a limited number of cache partitions, which causes degraded performance. These are major impediments to the practical adoption of software cache partitioning. In this paper, we propose a practical OS-level cache management scheme for multi-core real-time systems. Our scheme provides predictable cache performance, addresses the aforementioned problems of existing software cache partitioning, and efficiently allocates cache partitions to schedule a given task set. We have implemented and evaluated our scheme in Linux/RK running on the Intel Core i7 quad-core processor. Experimental results indicate that, compared to the traditional approaches, our scheme is up to 39% more memory space efficient and consumes up to 25% less cache partitions while maintaining cache predictability. Our scheme also yields a significant utilization benefit that increases with the number of tasks.

8 April 2013
Verified
Conference Paper
Open access
PDF file
JKJunsung Kim
Junsung Kim
HKHyoseung Kim
Hyoseung Kim
KLKarthik Lakshmanan
Karthik Lakshmanan
RRRagunathan (Raj) Rajkumar
Ragunathan (Raj) Rajkumar
As the complexity of software for Cyber-Physical Systems (CPS) rapidly increases, multi-core processors and parallel programming models such as OpenNP become appealing to CPS developers for guaranteeing timeliness. Hence, a parallel task on multi-core processors is expected to become a vital component in CPS such as a self-driving car, where tasks must be scheduled in real-time. In this paper, we extend the fork-join parallel task model to be scheduled in real-time, where the number of parallel threads can vary depending on the physical attributes of the system. To efficiently schedule the proposed task model, we develop the task stretch* transform. Using this transform for global Deadline Monotonic scheduling for fork-join real-time tasks, we achieve a resource augmentation bound of 3.73. In other words, any task set that is feasible on m unit-speed processors can be scheduled by the proposed algorithm on m processors that are 3.73 times faster. The proposed scheme is implemented on Linux/RK as a proof of concept, and ported to Boss, the self-driving vehicle that won the 2007 DARPA Urban Challenge. We evaluate our scheme on Boss by showing its driving quality, i.e., curvature and velocity profiles of the vehicle.

1 December 2013
Verified
Conference Paper
Closed access
NSNoriaki Suzuki
Noriaki Suzuki
HKHyoseung Kim
Hyoseung Kim
DNDionisio de Niz
Dionisio de Niz
BABjorn Andersson
Bjorn Andersson
LWLutz Wrage
Lutz Wrage
RRRagunathan Rajkumar
Ragunathan Rajkumar
In commercial-off-the-shelf (COTS) multi-core systems, the execution times of tasks become hard to predict because of contention on shared resources in the memory hierarchy. In particular, a task running in one processor core can delay the execution of another task running in another processor core. This is due to the fact that tasks can access data in the same cache set shared among processor cores or in the same memory bank in the DRAM memory (or both). Such cache and bank interference effects have motivated the need to create isolation mechanisms for resources accessed by more than one task. One popular isolation mechanism is cache coloring that divides the cache into multiple partitions. With cache coloring, each task can be assigned exclusive cache partitions, thereby preventing cache interference from other tasks. Similarly, bank coloring allows assigning exclusive bank partitions to tasks. While cache coloring and some bank coloring mechanisms have been studied separately, interactions between the two schemes have not been studied. Specifically, while memory accesses to two different bank colors do not interfere with each other at the bank level, they may interact at the cache level. Similarly, two different cache colors avoid cache interference but may not prevent bank interference. Therefore it is necessary to coordinate cache and bank coloring approaches. In this paper, we present a coordinated cache and bank coloring scheme that is designed to prevent cache and bank interference simultaneously. We also developed color allocation algorithms for configuring a virtual memory system to support our scheme which has been implemented in the Linux kernel. In our experiments, we observed that the execution time can increase by 60% due to inter-task interference when we use only cache coloring. Our coordinated approach can reduce this figure down to 12% (an 80% reduction).

23 February 2016 in Real-Time Systems
Verified
Journal Article
Closed access
PDF file
HKHyoseung Kim
Hyoseung Kim
DNDionisio de Niz
Dionisio de Niz
BABjörn Andersson
Björn Andersson
MKMark Klein
Mark Klein
OMOnur Mutlu
Onur Mutlu
RRRagunathan Rajkumar
Ragunathan Rajkumar