Understanding and optimizing persistent memory allocation

Cai, Wentao; Wen, Haosen; Beadle, H. Alan; Kjellqvist, Chris; Hedayati, Mohammad; Scott, Michael L.

doi:10.1145/3381898.3397212

Cited by 27 publications

(14 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Other persistent memory allocators also ask applications to replace raw pointers with similar designs of offset pointers (for example, libpmemobj library in PMDK [12] and Ralloc [15]). Furthermore, the concept of the non-raw pointer is being integrated into C++ (e.g., smart pointers).…”

Section: Pointers In Persistent Memorymentioning

confidence: 99%

“…Ralloc. Ralloc [15] is a persistent lock-free allocator designed and optimized for byte-addressable NVRAM (e.g., Intel Optane DC Persistent Memory).…”

Section: Implementations For Performance Comparisonmentioning

confidence: 99%

“…Ralloc showed notably better performance over libpmemobj in PMDK [15]. As Ralloc is targeted for byte-addressable NVRAM, we used it only on the Optane machine.…”

Section: Implementations For Performance Comparisonmentioning

confidence: 99%

“…nvm malloc [35] is another work for byte-addressable persistent memory with a fine-grained persistence policy. Ralloc [15] also targets byte-addressable persistent memory.…”

Section: Memory Allocatormentioning

confidence: 99%

“…It supports failure-atomic memory allocations by asking applications to write a function that traverses all active pointers. Ralloc possesses better performance over libpmemobj [15]. Compared with those works, Metall is designed for both block devices and byte-addressable persistent memory so that applications can utilize a wide range of persistent memory technologies in their environment.…”

Section: Memory Allocatormentioning

confidence: 99%

See 4 more Smart Citations

Metall: A Persistent Memory Allocator For Data-Centric Analytics

Iwabuchi,

Youssef,

Velusamy

et al. 2021

Preprint

View full text Add to dashboard Cite

Data analytics applications transform raw input data into analytics-specific data structures before performing analytics. Unfortunately, such data ingestion step is often more expensive than analytics. In addition, various types of NVRAM devices are already used in many HPC systems today. Such devices will be useful for storing and reusing data structures beyond a single process life cycle.We developed Metall, a persistent memory allocator built on top of the memory-mapped file mechanism. Metall enables applications to transparently allocate custom C++ data structures into various types of persistent memories. Metall incorporates a concise and high-performance memory management algorithm inspired by Supermalloc and the rich C++ interface developed by Boost.Interprocess library.On a dynamic graph construction workload, Metall achieved up to 11.7x and 48.3x performance improvements over Boost.Interprocess and memkind (PMEM kind), respectively. We also demonstrate Metall's high adaptability by integrating Metall into a graph processing framework, GraphBLAS Template Library. This study's outcomes indicate that Metall will be a strong tool for accelerating future large-scale data analytics by allowing applications to leverage persistent memory efficiently.

show abstract

Section: Pointers In Persistent Memorymentioning

confidence: 99%

“…Ralloc. Ralloc [15] is a persistent lock-free allocator designed and optimized for byte-addressable NVRAM (e.g., Intel Optane DC Persistent Memory).…”

Section: Implementations For Performance Comparisonmentioning

confidence: 99%

“…Ralloc showed notably better performance over libpmemobj in PMDK [15]. As Ralloc is targeted for byte-addressable NVRAM, we used it only on the Optane machine.…”

Section: Implementations For Performance Comparisonmentioning

confidence: 99%

“…nvm malloc [35] is another work for byte-addressable persistent memory with a fine-grained persistence policy. Ralloc [15] also targets byte-addressable persistent memory.…”

Section: Memory Allocatormentioning

confidence: 99%

Section: Memory Allocatormentioning

confidence: 99%

See 3 more Smart Citations