A compiler framework for general memory layout optimizations targeting structures

Lin, Jin; Yew, Pen-Chung

doi:10.1145/1739025.1739033

Cited by 8 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because this metric captures access locality of two members of a structure, it can also be used to estimate how harmful to split two members into separate memory regions. Other studies such as [18,21] propose similar metrics. Applying these metrics is easier than implementing data partitioning because they are based on memory traces of the original program.…”

Section: Analyzing Performance Implicationmentioning

confidence: 87%

“…For the second part, no off-the-shelf tool that enables data partitioning is available even though much work have been done on structure splitting, mostly because implementing structure splitting for general cases is very difficult. Instead, we can use existing metrics [18,21,31] that predict the benefit of structure splitting for a given structure.…”

Section: Discussionmentioning

confidence: 99%

“…This can dramatically degrade the performance due to reduced access locality. struct tree_node { int id; // id of the node, critical struct tree_node * r; // pointer to the right child, critical struct tree_node * l; // pointer to the left child, critical double score; // score of this node, approximate }; int size = 1000 * sizeof(struct tree_node); struct tree_node * nodes = malloc(size); Prior works [8,18,31,33] propose to convert "an array of structures" to "a structure of arrays" for improving access locality. For example, the code in Figure 4 can be converted to have a distinct array for each member of struct tree_node, and then the array for score can be mapped to a consecutive memory region with high error rate.…”

Section: Challenge: Interleaved Criticalitymentioning

confidence: 99%

See 2 more Smart Citations

Assessing Impact of Data Partitioning for Approximate Memory in C/C++ Code

Akiyama¹

2020

Preprint

View full text Add to dashboard Cite

Approximate memory is a technique to mitigate the performance gap between memory subsystems and CPUs with its reduced access latency at a cost of data integrity. To gain benefit from approximate memory for realistic applications, it is crucial to partition applications' data to approximate data and critical data and apply different error rates. However, error rates cannot be controlled in a fine-grained manner (e.g., per byte) due to fundamental limitations of how approximate memory can be realized. Due to this, if approximate data and critical data are interleaved in a data structure (e.g., a C struct that has a pointer and an approximatable number as its members), data partitioning may degrade the application's performance because the data structure must be split to separate memory regions that have different error rates. This paper is the first to conduct an analysis of realistic C/C++ code to assess the impact of this problem. First, we find the type of data (e.g., "int", "struct point") that is assessed by the instruction that incurs the largest number of cache misses in a benchmark, which we refer to as the target data type. Second, we qualitatively estimate if the target data type of an application has approximate data and critical data interleaved. To this end, we set up three criteria to analyze it because definitively distinguishing a piece of data as approximate data or critical data is infeasible since it depends on each use-case. We analyze 11 memory intensive benchmarks from SPEC CPU 2006 and 2 graph analytics frameworks, and show that the target data types of 9 benchmarks are either a C struct or a C++ class (criterion 1). Among them, two have a pointer and a non-pointer member together (criterion 2) and three have a floating point number and other members together (criterion 3).

show abstract

Section: Analyzing Performance Implicationmentioning

confidence: 87%

Section: Discussionmentioning

confidence: 99%

Section: Challenge: Interleaved Criticalitymentioning

confidence: 99%

See 1 more Smart Citation

Assessing Impact of Data Partitioning for Approximate Memory in C/C++ Code

Akiyama¹

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Alias group-wise safety analysis [Lin and Yew 2010] can identify a finer-grained splitting candidate. The violation from one instance will only affect the alias group it belongs to, not the entire type, so more opportunities can be exposed.…”

Section: Discussionmentioning

confidence: 99%

“…They also proposed an advisory tool that combined static analysis with runtime profiling to guide structure layout decisions. Lin and Yew [2010] proposed to use groups of objects as candidates, instead of objects of the same type. The objects in the same group are aliased with each other whereas the objects in different groups are not aliased.…”

Section: Related Workmentioning

confidence: 99%

On-the-fly structure splitting for heap objects

Wang¹,

Wu²,

Yew

et al. 2012

ACM Trans. Archit. Code Optim.

Self Cite

View full text Add to dashboard Cite

With the advent of multicore systems, the gap between processor speed and memory latency has grown worse because of their complex interconnect. Sophisticated techniques are needed more than ever to improve an application's spatial and temporal locality. This paper describes an optimization that aims to improve heap data layout by structure-splitting. It also provides runtime address checking by piggybacking on the existing page protection mechanism to guarantee the correctness of such optimization that has eluded many previous attempts due to safety concerns. The technique can be applied to both sequential and parallel programs at either compile time or runtime. However, we focus primarily on sequential programs (i.e., single-threaded programs) at runtime in this paper. Experimental results show that some benchmarks in SPEC 2000 and 2006 can achieve a speedup of up to 142.8%.

show abstract

Data layout optimization based on the spatio-temporal model of field access

Wang

et al. 2022

2022 5th International Conference on Advanced Electronic Materials, Computers and Software Engineering (AEMCSE)

View full text Add to dashboard Cite

A compiler framework for general memory layout optimizations targeting structures

Cited by 8 publications

References 13 publications

Assessing Impact of Data Partitioning for Approximate Memory in C/C++ Code

Assessing Impact of Data Partitioning for Approximate Memory in C/C++ Code

On-the-fly structure splitting for heap objects

Data layout optimization based on the spatio-temporal model of field access

Contact Info

Product

Resources

About