Vijay S. Pai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Vijay S. Pai has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Yung Ryn Choe
Purdue University, West Lafayette, IN 47907
3 shared publications - Derek L. Schuff
Purdue University, West Lafayette, IN 47907
2 shared publications - Abhisek Pan
Purdue University, West Lafayette, IN
1 shared publication - Chase Douglas
Purdue University, West Lafayette, IN 47907
1 shared publication - David I. August
Princeton University, Princeton, New Jersey, USA
1 shared publication - Li-Shiuan Peh
Princeton University
1 shared publication
Latest Publications
- Open Access
- 15 November 2015
- Published by Association for Computing Machinery (ACM)
- p. 1-12
Runtime-driven shared last-level cache management for task-parallel programs
Task-parallel programming models with input annotation-based concurrency extraction at runtime present a promising paradigm for programming multicore processors. Through management of dependencies, task assignments, and orchestration, these models markedly simplify the programming effort for parallelization while exposing higher levels of concurrency. In this paper we show that for multicores with a shared last-level cache (LLC), the concurrency extraction framework can be used to improve the shared LLC performance. Based on the input annotations for future tasks, the runtime instructs the hardware to prioritize data blocks with future reuse while evicting blocks with no future reuse. These instructions allow the hardware to preserve all the blocks for at least some of the future tasks and evict dead blocks. This leads to a considerable improvement in cache efficiency over what is achieved by hardware-only replacement policies, which can replace blocks for all future tasks resulting in poor hit-rates for all future tasks. The proposed hardware-software technique leads to a mean improvement of 18% in application performance and a mean reduction of 26% in misses over a shared LLC managed by the Least Recently Used replacement policy for a set of input-annotated task-parallel programs using the OmpSs programming model implemented on the NANOS++ runtime. In contrast, the state-of-the-art thread-based partitioning scheme suffers an average performance loss of 2% and an average increase of 15% in misses over the baseline.
- 15 November 2015
- Published by Association for Computing Machinery (ACM)
- p. 1-12
Open AccessRuntime-driven shared last-level cache management for task-parallel programs
Task-parallel programming models with input annotation-based concurrency extraction at runtime present a promising paradigm for programming multicore processors. Through management of dependencies, task assignments, and orchestration, these models markedly simplify the programming effort for parallelization while exposing higher levels of concurrency. In this paper we show that for multicores with a shared last-level cache (LLC), the concurrency extraction framework can be used to improve the shared LLC performance. Based on the input annotations for future tasks, the runtime instructs the hardware to prioritize data blocks with future reuse while evicting blocks with no future reuse. These instructions allow the hardware to preserve all the blocks for at least some of the future tasks and evict dead blocks. This leads to a considerable improvement in cache efficiency over what is achieved by hardware-only replacement policies, which can replace blocks for all future tasks resulting in poor hit-rates for all future tasks. The proposed hardware-software technique leads to a mean improvement of 18% in application performance and a mean reduction of 26% in misses over a shared LLC managed by the Least Recently Used replacement policy for a set of input-annotated task-parallel programs using the OmpSs programming model implemented on the NANOS++ runtime. In contrast, the state-of-the-art thread-based partitioning scheme suffers an average performance loss of 2% and an average increase of 15% in misses over the baseline.
- 1 April 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 33, 32-43
Conservative vs. Optimistic Parallelization of Stateful Network Intrusion Detection
This paper presents and experimentally analyzes the performance of three parallelization strategies for the popular open-source Snort network intrusion detection system (NIDS). The parallelizations include 2 conservative variants and 1 optimistic scheme. The conservative strategy parallelizes inspection at the level of TCP/IP flows, as any potential inter-packet dependences are confined to a single flow. The flows are partitioned among threads, and each flow is processed in-order at one thread. A second variation reassigns flows between threads to improve load balance but still requires that only one thread process a given flow at a time. The flow-concurrent scheme provides good performance for 3 of the 5 network packet traces studied, reaching as high as 4.1 speedup and 3.1 Gbps inspection rate on a commodity 8-core server. Dynamic reassignment does not improve performance scalability because it introduces locking overheads that offset any potential benefits of load balancing. Neither conservative version can achieve good performance, however, without enough concurrent networkflows. For this case, this paper presents an optimistic parallelization that exploits the observation that not all packets from a flow are actually connected by dependences. This system allows a single flow to be simultaneously processed by multiple threads, stalling if an actual dependence is found. The optimistic version has additional overheads that reduce speedup by 25% for traces with flow concurrency, but its benefits allow one additional trace to see substantial speedup (2.4 on five cores).
- 1 April 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 33, 32-43
Conservative vs. Optimistic Parallelization of Stateful Network Intrusion Detection
This paper presents and experimentally analyzes the performance of three parallelization strategies for the popular open-source Snort network intrusion detection system (NIDS). The parallelizations include 2 conservative variants and 1 optimistic scheme. The conservative strategy parallelizes inspection at the level of TCP/IP flows, as any potential inter-packet dependences are confined to a single flow. The flows are partitioned among threads, and each flow is processed in-order at one thread. A second variation reassigns flows between threads to improve load balance but still requires that only one thread process a given flow at a time. The flow-concurrent scheme provides good performance for 3 of the 5 network packet traces studied, reaching as high as 4.1 speedup and 3.1 Gbps inspection rate on a commodity 8-core server. Dynamic reassignment does not improve performance scalability because it introduces locking overheads that offset any potential benefits of load balancing. Neither conservative version can achieve good performance, however, without enough concurrent networkflows. For this case, this paper presents an optimistic parallelization that exploits the observation that not all packets from a flow are actually connected by dependences. This system allows a single flow to be simultaneously processed by multiple threads, stalling if an actual dependence is found. The optimistic version has additional overheads that reduce speedup by 25% for traces with flow concurrency, but its benefits allow one additional trace to see substantial speedup (2.4 on five cores).
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15302075,p. 1-10
Achieving Reliable Parallel Performance in a VoD Storage Server Using Randomization and Replication
This paper investigates randomization and replication as strategies to achieve reliable performance in disk arrays targeted for video-on-demand (VoD) workloads. A disk array can provide high aggregate throughput, but only if the server can effectively balance the load on the disks. Such load balance is complicated by two key factors: workload hotspots caused by differences in popularity among media streams, and "fail-stutter" faults that arise when the performance of one or more devices drops below expectations due to manufacturing variations, hardware problems, or geometry-related variations. This paper focuses on the random duplicate assignment (RDA) data allocation policy which places each data block on two disks chosen at random, independent of other blocks in the same media stream or other streams. This strategy is compared to traditional single-disk file allocation, disk striping (RAID-0), disk mirroring (RAID-1), and randomization without duplication. The various allocation schemes are implemented and tested using a prototype VoD server with 2 dual-core Opteron processors, 8 SATA disks, and 4 gigabit Ethernet interfaces running the Linux 2.6 kernel. The results indicate that combining randomization and replication allows RDA to effectively tolerate both workload hotspots and fail-stutter faults better than previous schemes.
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15302075,p. 1-10
Achieving Reliable Parallel Performance in a VoD Storage Server Using Randomization and Replication
This paper investigates randomization and replication as strategies to achieve reliable performance in disk arrays targeted for video-on-demand (VoD) workloads. A disk array can provide high aggregate throughput, but only if the server can effectively balance the load on the disks. Such load balance is complicated by two key factors: workload hotspots caused by differences in popularity among media streams, and "fail-stutter" faults that arise when the performance of one or more devices drops below expectations due to manufacturing variations, hardware problems, or geometry-related variations. This paper focuses on the random duplicate assignment (RDA) data allocation policy which places each data block on two disks chosen at random, independent of other blocks in the same media stream or other streams. This strategy is compared to traditional single-disk file allocation, disk striping (RAID-0), disk mirroring (RAID-1), and randomization without duplication. The various allocation schemes are implemented and tested using a prototype VoD server with 2 dual-core Opteron processors, 8 SATA disks, and 4 gigabit Ethernet interfaces running the Linux 2.6 kernel. The results indicate that combining randomization and replication allows RDA to effectively tolerate both workload hotspots and fail-stutter faults better than previous schemes.
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Design Alternatives for a High-Performance Self-Securing Ethernet Network Interface
This paper presents and evaluates a strategy for integrating the Snort network intrusion detection system into a high-performance programmable Ethernet network interface card (NIC), considering the impact of several possible hardware and software design choices. While currently proposed ASIC, FPGA, and TCAM systems can match incoming string content in real-time, the system proposed also supports the stream reassembly and HTTP content transformation capabilities of Snort. This system, called LineSnort, parallelizes Snort using concurrency across TCP sessions and executes those parallel tasks on multiple low-frequency pipelined RISC processors embedded in the NIC. LineSnort additionally exploits opportunities for intra-session concurrency. The system also includes dedicated hardware for high-bandwidth data transfers and for high-performance string matching.
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Design Alternatives for a High-Performance Self-Securing Ethernet Network Interface
This paper presents and evaluates a strategy for integrating the Snort network intrusion detection system into a high-performance programmable Ethernet network interface card (NIC), considering the impact of several possible hardware and software design choices. While currently proposed ASIC, FPGA, and TCAM systems can match incoming string content in real-time, the system proposed also supports the stream reassembly and HTTP content transformation capabilities of Snort. This system, called LineSnort, parallelizes Snort using concurrency across TCP sessions and executes those parallel tasks on multiple low-frequency pipelined RISC processors embedded in the NIC. LineSnort additionally exploits opportunities for intra-session concurrency. The system also includes dedicated hardware for high-bandwidth data transfers and for high-performance string matching.
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15302075,p. 1-7
A Model and Prototype of a Resource-Efficient Storage Server for High-Bitrate Video-on-Demand
This paper presents a mathematical model and a prototype of a resource-efficient storage server for high-bitrate video-on-demand (VoD) applications. Rapid exponential growth of disk capacity enables the storage of high-bitrate VoD streams; however, a server system must be carefully designed to allow those streams to be retrieved from disk and delivered to the network efficiently. Additionally, a cost-effective server should be implemented using only commodity components, such as standard PCs, SATA disks and controllers, and Gigabit Ethernet links. This paper presents a model: detailed enough to account for the rate-based nature of streaming video, the buffering time allowed by the application, and average-case disk hardware characteristics while remaining simple enough to use for algorithm and system design. This paper then describes a prototype storage server designed to serve large video files at the specified bitrates and finds its performance to agree closely with the model (with an average discrepancy of 11% for high-bitrate streams). The system uses up to 8 SATA-300 disks and can simultaneously serve 290 distinct DVD-quality (6 Mbps) streams or 74 distinct HDTV-quality (25 Mbps) streams from disk, achieving an aggregate network throughput of 1.85 Gbps.
- 1 January 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15302075,p. 1-7
A Model and Prototype of a Resource-Efficient Storage Server for High-Bitrate Video-on-Demand
This paper presents a mathematical model and a prototype of a resource-efficient storage server for high-bitrate video-on-demand (VoD) applications. Rapid exponential growth of disk capacity enables the storage of high-bitrate VoD streams; however, a server system must be carefully designed to allow those streams to be retrieved from disk and delivered to the network efficiently. Additionally, a cost-effective server should be implemented using only commodity components, such as standard PCs, SATA disks and controllers, and Gigabit Ethernet links. This paper presents a model: detailed enough to account for the rate-based nature of streaming video, the buffering time allowed by the application, and average-case disk hardware characteristics while remaining simple enough to use for algorithm and system design. This paper then describes a prototype storage server designed to serve large video files at the specified bitrates and finds its performance to agree closely with the model (with an average discrepancy of 11% for high-bitrate streams). The system uses up to 8 SATA-300 disks and can simultaneously serve 290 distinct DVD-quality (6 Mbps) streams or 74 distinct HDTV-quality (25 Mbps) streams from disk, achieving an aggregate network throughput of 1.85 Gbps.
- 1 June 2005
- Published by Springer Nature in International Journal of Parallel Programming
- Vol. 33 (2-3) , 81-101
Achieving Structural and Composable Modeling of Complex Systems
- 1 June 2005
- Published by Springer Nature in International Journal of Parallel Programming
- Vol. 33 (2-3) , 81-101
Achieving Structural and Composable Modeling of Complex Systems