Flexible Hardware Acceleration for Instruction-Grain Lifeguards

Chen, S.; Kozuch, Michael A.; Gibbons, Phillip B.; Ryan, Michael P.; Strigkos, Theodoros; Mowry, Todd C.; Ruwase, Olatunji; Vlachos, Evangelos; Falsafi, Babak; Ramachandran, Vijaya

doi:10.1109/mm.2009.6

Cited by 10 publications

(14 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instruction-Grain Lifeguards. Instruction-grain lifeguards often have common structures, as observed by previous works [5,6,30,31]. First, a lifeguard often maintains a piece of state (called metadata or shadow state) for every memory location (and/or register) in the application.…”

Section: Background: Online Sequential Monitoringmentioning

confidence: 96%

“…As we observe later in our experimental results, the performance of this state-of-the-art approach is unacceptably poor because neither the application nor the lifeguard can enjoy parallel speedups. To enable both the application and the lifeguard to operate in parallel with high performance, we must overcome three key challenges: (i) adapting the three hardware accelerators from our earlier study [5,6] to work for multithreaded applications, (ii) capturing inter-thread data dependences between the application threads and reproducing their effects appropriately in the lifeguard processing, and (iii) ensuring the multiple lifeguard threads atomically access their shared view (called metadata) of the application's state at negligible cost. Parallelizing Lifeguard Hardware Accelerators.…”

Section: Challenges In Online Parallel Monitoring Of Multithreaded Apmentioning

confidence: 99%

“…Parallelizing Lifeguard Hardware Accelerators. The three lifeguard accelerators described in our earlier study [5,6] are inheritance tracking, idempotency filters, and metadata-TLBs. These accelerators, which are applicable to stream-based lifeguard frameworks such as LBA [4] and DISE [8], are a key enabler for online, instruction-grain monitoring.…”

Section: Challenges In Online Parallel Monitoring Of Multithreaded Apmentioning

confidence: 99%

“…• We demonstrate how the three hardware accelerators that we proposed earlier [5,6] as key enablers of online single-threaded monitoring can be modified to support online multithreaded monitoring. We present our system design (including its accelerators) for both Sequentially Consistent and Total Store Ordering processors.…”

Section: Contributionsmentioning

confidence: 99%

“…Fortunately, we recently demonstrated [5,6] that in the case of monitoring singlethreaded applications, a set of hardware-based lifeguard accelerators can dramatically improve performance to the point where the remaining overheads are relatively modest. Namely, the overhead drops from 3-5X to 1.02-1.5X for a number of diverse lifeguards.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

ParaLog

Vlachos¹,

Goodstein²,

Kozuch³

et al. 2010

Proceedings of the Fifteenth International Conference on Architectural Support for Programming Languages and Operating Systems

Self Cite

View full text Add to dashboard Cite

Instruction-grain lifeguards monitor the events of a running application at the level of individual instructions in order to identify and help mitigate application bugs and security exploits. Because such lifeguards impose a 10-100X slowdown on existing platforms, previous studies have proposed hardware designs to accelerate lifeguard processing. However, these accelerators are either tailored to a specific class of lifeguards or suitable only for monitoring singlethreaded programs.We present ParaLog, the first design of a system enabling fast online parallel monitoring of multithreaded parallel applications. ParaLog supports a broad class of software-defined lifeguards. We show how three existing accelerators can be enhanced to support online multithreaded monitoring, dramatically reducing lifeguard overheads. We identify and solve several challenges in monitoring parallel applications and/or parallelizing these accelerators, including (i) enforcing inter-thread data dependences, (ii) dealing with inter-thread effects that are not reflected in coherence traffic, (iii) dealing with unmonitored operating system activity, and (iv) ensuring lifeguards can access shared metadata with negligible synchronization overheads. We present our system design for both Sequentially Consistent and Total Store Ordering processors. We implement and evaluate our design on a 16 core simulated CMP, using benchmarks from SPLASH-2 and PARSEC and two lifeguards: a data-flow tracking lifeguard and a memory-access checker lifeguard. Our results show that (i) our parallel accelerators improve performance by 2-9X and 1.13-3.4X for our two lifeguards, respectively, (ii) we are 5-126X faster than the time-slicing approach required by existing techniques, and (iii) our average overheads for applications with eight threads are 51% and 28% for the two lifeguards, respectively.

show abstract

Section: Background: Online Sequential Monitoringmentioning

confidence: 96%

Section: Challenges In Online Parallel Monitoring Of Multithreaded Apmentioning

confidence: 99%

Section: Challenges In Online Parallel Monitoring Of Multithreaded Apmentioning

confidence: 99%

Section: Contributionsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

ParaLog

Vlachos¹,

Goodstein²,

Kozuch³

et al. 2010

Proceedings of the Fifteenth International Conference on Architectural Support for Programming Languages and Operating Systems

Self Cite

View full text Add to dashboard Cite

show abstract

Efficient dynamic program monitoring on multi-core systems

Zhai

2011

Journal of Systems Architecture

View full text Add to dashboard Cite

ParaLog

Vlachos

Goodstein

Kozuch

et al. 2010

SIGARCH Comput. Archit. News

View full text Add to dashboard Cite

Instruction-grain lifeguards monitor the events of a running application at the level of individual instructions in order to identify and help mitigate application bugs and security exploits. Because such lifeguards impose a 10-100X slowdown on existing platforms, previous studies have proposed hardware designs to accelerate lifeguard processing. However, these accelerators are either tailored to a specific class of lifeguards or suitable only for monitoring singlethreaded programs. We present ParaLog, the first design of a system enabling fast online parallel monitoring of multithreaded parallel applications. ParaLog supports a broad class of software-defined lifeguards. We show how three existing accelerators can be enhanced to support online multithreaded monitoring, dramatically reducing lifeguard overheads. We identify and solve several challenges in monitoring parallel applications and/or parallelizing these accelerators, including (i) enforcing inter-thread data dependences, (ii) dealing with inter-thread effects that are not reflected in coherence traffic, (iii) dealing with unmonitored operating system activity, and (iv) ensuring lifeguards can access shared metadata with negligible synchronization overheads. We present our system design for both Sequentially Consistent and Total Store Ordering processors. We implement and evaluate our design on a 16 core simulated CMP, using benchmarks from SPLASH-2 and PARSEC and two lifeguards: a data-flow tracking lifeguard and a memory-access checker lifeguard. Our results show that (i) our parallel accelerators improve performance by 2-9X and 1.13-3.4X for our two lifeguards, respectively, (ii) we are 5-126X faster than the time-slicing approach required by existing techniques, and (iii) our average overheads for applications with eight threads are 51% and 28% for the two lifeguards, respectively.

show abstract

Flexible Hardware Acceleration for Instruction-Grain Lifeguards

Cited by 10 publications

References 14 publications

ParaLog

ParaLog

Efficient dynamic program monitoring on multi-core systems

ParaLog

Contact Info

Product

Resources

About