MemGuard: Memory bandwidth reservation system for efficient performance isolation in multi-core platforms

@article{Yun2013MemGuardMB,
  title={MemGuard: Memory bandwidth reservation system for efficient performance isolation in multi-core platforms},
  author={Heechul Yun and Gang Yao and Rodolfo Pellizzoni and Marco Caccamo and Lui Raymond Sha},
  journal={2013 IEEE 19th Real-Time and Embedded Technology and Applications Symposium (RTAS)},
  year={2013},
  pages={55-64}
}
  • H. Yun, Gang Yao, +2 authors L. Sha
  • Published 9 April 2013
  • Computer Science
  • 2013 IEEE 19th Real-Time and Embedded Technology and Applications Symposium (RTAS)
Memory bandwidth in modern multi-core platforms is highly variable for many reasons and is a big challenge in designing real-time systems as applications are increasingly becoming more memory intensive. In this work, we proposed, designed, and implemented an efficient memory bandwidth reservation system, that we call MemGuard. MemGuard distinguishes memory bandwidth as two parts: guaranteed and best effort. It provides bandwidth reservation for the guaranteed bandwidth for temporal isolation… Expand
Memory Bandwidth Management for Efficient Performance Isolation in Multi-Core Platforms
TLDR
MemGuard separates memory bandwidth in two parts: guaranteed and best effort, and provides bandwidth reservation for the guaranteed bandwidth for temporal isolation, with efficient reclaiming to maximally utilize the reserved bandwidth. Expand
Memguard, Memory Bandwidth Management in Mixed Criticality Virtualized Systems - Memguard KVM Scheduling
TLDR
A new implementation of Memguard is proposed that can also be utilized in mixed-criticality virtualized computing environments and includes modifications to the CFS (Completely Faire Scheduler) Linux scheduler to work with the modified Memguard kernel module. Expand
Improving Real-Time Performance on Multicore Platforms Using MemGuard
In this paper, we present a case-study that shows the challen ges of using multicore based platforms for real-time applications, and demonstrate how our kernel based memory b andwidth managementExpand
BRU: Bandwidth Regulation Unit for Real-Time Multicore Processors
TLDR
The Bandwidth Regulation Unit is presented, a drop-in hardware module that enables per-core memory bandwidth regulation at fine-grained time intervals and has the ability to regulate the memory access bandwidth of multiple cores collectively to improve bandwidth utilization. Expand
PALLOC: DRAM bank-aware memory allocator for performance isolation on multicore platforms
TLDR
PALLOC, a DRAM bank-aware memory allocator which exploits the page-based virtual memory system to allocate memory pages of each application to specific banks, thereby improving isolation on COTS multicore platforms without requiring any special hardware support. Expand
Protecting Memory-Performance Critical Sections in Soft Real-Time Applications
TLDR
This paper finds BWLOCK, user-level APIs and a memory bandwidth control mechanism that can protect such memory performance critical sections of soft real-time applications do exist and are often easy to identify; and applying BWLOCK for memory critical sections significantly improve performance of the softRealTime applications at a small or no cost in throughput of non real- time applications. Expand
Dynamic Memory Bandwidth Allocation for Real-Time GPU-Based SoC Platforms
TLDR
This article proposes a novel memory bandwidth allocation scheme where it dynamically monitor the progress of a real-time application and increase the bandwidth share of BE ones whenever it is safe to do so, and demonstrates its effectiveness on a variety of GPU and CPU benchmarks. Expand
BWLOCK: A Dynamic Memory Access Control Framework for Soft Real-Time Applications on Multicore Platforms
TLDR
This paper presents a memory access control framework called BWLOCK, which is designed to protect MPCSs of soft real-time applications, and achieves significant performance improvements, up to 150 percent reduction in slowdown, at a controllable throughput impact to non real- time applications. Expand
EMBA: Efficient Memory Bandwidth Allocation to Improve Performance on Intel Commodity Processor
TLDR
This work forms a quantitative relationship between a program's performance and its LLC occupancy and memory request rate on commodity processors, and proposes a heuristic bound-aware throttling algorithm to improve system performance and develops a hierarchical clustering method to improve the algorithm's efficiency. Expand
Memory-Aware Fair-Share Scheduling for Improved Performance Isolation in the Linux Kernel
TLDR
A memory-aware fair-share scheduling algorithm that can make QoS applications less susceptible to memory-related interference from other co-running applications and is a software-only solution that can be easily integrated into the kernel scheduler with only minimal modification to the kernel. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 23 REFERENCES
Memory Access Control in Multiprocessor for Real-Time Systems with Mixed Criticality
TLDR
This paper developed analytic solutions to compute proper throttling parameters that satisfy schedulability of critical tasks while minimize performance impact caused by throttling and implemented the mechanism in Linux kernel. Expand
Process Cruise Control: Throttling Memory Access in a Soft Real-Time Environment
TLDR
This work demonstrates the benefit of the memory throttle for a video conferencing application running in a multiprogrammed multiprocessor environment and introduces memory-bandwidth guarantees, which adds a further resource to capacity reservation models and therefore enhances the quality of service. Expand
PRET DRAM controller: Bank privatization for predictability and temporal isolation
TLDR
This work introduces a novel DRAM controller design that is predictable and that significantly reduces worst-case access latencies, and its integration with a precision-timed (PRET) architecture called PTARM. Expand
Predator: A predictable SDRAM memory controller
  • B. Akesson, K. Goossens, M. Ringhofer
  • Computer Science
  • 2007 5th IEEE/ACM/IFIP International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)
  • 2007
TLDR
A memory controller design that provides a guaranteed minimum bandwidth and a maximum latency bound to the IPs is presented using a novel two-step approach to predictable SDRAM sharing. Expand
Fair Queuing Memory Systems
TLDR
The proposed memory scheduler is fair and provides quality of service (QoS) while improving system performance and reduces the variance in the threads' target memory bandwidth utilization. Expand
Efficient reclaiming in reservation-based real-time systems with variable execution times
TLDR
It is proven that the reclaiming mechanism does not violate the isolation property of the CBS and can be safely adopted to achieve temporal protection even when resource reservations are not precisely assigned. Expand
Real-Time Scheduling Using Credit-Controlled Static-Priority Arbitration
TLDR
It is shown that CCSP belongs to the class of latency-rate servers and guarantees the allocated rate within a maximum latency, as required by hard real-time applications, and a hardware implementation of the arbiter is presented in the context of a DDR2 SDRAM controller. Expand
Greedy reclamation of unused bandwidth in constant-bandwidth servers
  • G. Lipari, Sanjoy Baruah
  • Computer Science
  • Proceedings 12th Euromicro Conference on Real-Time Systems. Euromicro RTS 2000
  • 2000
A framework for scheduling a number of different applications on a single shared pre-emptable processor is proposed, such that each application seems to be executing on a slower dedicated processor.Expand
Capacity sharing for overrun control
TLDR
The proposed method achieves isolation among tasks through a resource reservation mechanism which bounds the effects of task interference but which also performs efficient reclamation of the unused computation times in order to relax the utilization constraints imposed by isolation. Expand
TimeGraph: GPU Scheduling for Real-Time Multi-Tasking Environments
TLDR
TimeGraph is presented, a real-time GPU scheduler at the device-driver level for protecting important GPU workloads from performance interference and supports two priority-based scheduling policies in order to address the tradeoff between response times and throughput introduced by the asynchronous and non-preemptive nature of GPU processing. Expand
...
1
2
3
...