Rethinking software runtimes for disaggregated memory

Calciu, Irina; Imran, Muhammad; Puddu, Ivan; Kashyap, Sanidhya; Maruf, Hasan Al; Mutlu, Onur; Kolli, Aasheesh

doi:10.1145/3445814.3446713

Cited by 77 publications

(43 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…given the limited time and money available for artifact evaluation, we advise reviewers not to attempt to manufacture their own Enzians, and instead to proceed with the second artifact, found at Zenodo [21]. The provided bitstreams and benchmarks will allow reviewers to reproduce all presented evaluation data visualized in Figures 5,6,7,8,10,11…”

Section: Discussionmentioning

confidence: 99%

“…Dagger [39] implements Remote Procedure Call (RPC) on the FPGA to use it as a smart NIC, taking advantage of the network connection available in the Arria 10 FPGA. Kona [8] uses the coherent interconnect to implement remote memory by exposing łfakež physical memory backed by memory on other Kona machines.…”

Section: The Changing Fpga Platform Landscapementioning

confidence: 99%

“…[18]) integrate CPU cores onto an FPGA die. Access to the cache protocol on the chip in these systems allows a use-case not possible with Intel HARP: the FPGA acting as part of the CPU's memory system, enabling research into new remote memory protocols [8,9] and cache management [58]. However, such systems are tiny in comparison with server machines, typically using 4 low-power ARM cores with limited DRAM.…”

Section: The Changing Fpga Platform Landscapementioning

confidence: 99%

See 2 more Smart Citations

Enzian: an open, general, CPU/FPGA platform for systems software research

Cock

Ramdas

Schwyn

et al. 2022

Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

Section: Discussionmentioning

confidence: 99%

Section: The Changing Fpga Platform Landscapementioning

confidence: 99%

Section: The Changing Fpga Platform Landscapementioning

confidence: 99%

See 1 more Smart Citation

Enzian: an open, general, CPU/FPGA platform for systems software research

Cock

Ramdas

Schwyn

et al. 2022

Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

“…While smaller granularities of transfer can moderate extraneous data movement in a few applications, the overhead of initiating fine-grain transfers from the CPU remains high enough to nullify any scope for improvement [61]. In general, applications with input-dependent irregular memory accesses can potentially incur extraneous data movement [2,12,14]. Direct access to PM: An obvious benefit of GPM is its ability to manipulate PM-resident data structures using direct loads and stores from the GPU.…”

Section: Understanding the Benefits Of Gpmmentioning

confidence: 99%

“…Third, sometimes only a fraction of data is updated during computation. However, which data would be updated is not known apriori [2,12,14]. Since the GPU cannot directly persist results while computing, extraneous data could be transferred to and persisted by the CPU.…”

Section: Introductionmentioning

confidence: 99%

GPM: leveraging persistent memory from a GPU

Pandey¹,

Kamath²,

Basu³

2022

Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

The GPU is a key computing platform for many application domains. While the new non-volatile memory technology has brought the promise of byte-addressable persistence (a.k.a., persistent memory, or PM) to CPU applications, the same, unfortunately, is beyond the reach of GPU programs.We take three key steps toward enabling GPU programs to access PM directly. First, enable direct access to PM from within a GPU kernel without needing to modify the hardware. Next, we demonstrate three classes of GPU-accelerated applications that benefit from PM. In the process, we create a workload suite with nine such applications. We then create a GPU library, written in CUDA, to support logging, checkpointing, and primitives for native persistence for programmers to easily leverage PM. CCS CONCEPTS• Computer systems organization → Processors and memory architectures.

show abstract

Evaluating the potential of disaggregated memory systems for HPC applications

Ding,

Maris,

Nam

et al. 2024

Concurrency and Computation

View full text Add to dashboard Cite

SummaryDisaggregated memory is a promising approach that addresses the limitations of traditional memory architectures by enabling memory to be decoupled from compute nodes and shared across a data center. Cloud platforms have deployed such systems to improve overall system memory utilization, but performance can vary across workloads. High‐performance computing (HPC) is crucial in scientific and engineering applications, where HPC machines also face the issue of underutilized memory. As a result, improving system memory utilization while understanding workload performance is essential for HPC operators. Therefore, learning the potential of a disaggregated memory system before deployment is a critical step. This paper proposes a methodology for exploring the design space of a disaggregated memory system. It incorporates key metrics that affect performance on disaggregated memory systems: memory capacity, local and remote memory access ratio, injection bandwidth, and bisection bandwidth, providing an intuitive approach to guide machine configurations based on technology trends and workload characteristics. We apply our methodology to analyze thirteen diverse workloads, including AI training, data analysis, genomics, protein, fusion, atomic nuclei, and traditional HPC bookends. Our methodology demonstrates the ability to comprehend the potential and pitfalls of a disaggregated memory system and provides motivation for machine configurations. Our results show that eleven of our thirteen applications can leverage injection bandwidth disaggregated memory without affecting performance, while one pays a rack bisection bandwidth penalty and two pay the system‐wide bisection bandwidth penalty. In addition, we also show that intra‐rack memory disaggregation would meet the application's memory requirement and provide enough remote memory bandwidth.

show abstract

Rethinking software runtimes for disaggregated memory

Cited by 77 publications

References 61 publications

Enzian: an open, general, CPU/FPGA platform for systems software research

Enzian: an open, general, CPU/FPGA platform for systems software research

GPM: leveraging persistent memory from a GPU

Evaluating the potential of disaggregated memory systems for HPC applications

Contact Info

Product

Resources

About