Towards analysis and refinement of auto-tuning spaces

J 2026

Towards analysis and refinement of auto-tuning spaces

FILIPOVIČ, Jiří; Suren Harutyunyan GEVORGYAN; Eduardo CESAR a Anna SIKORA

Základní údaje

Originální název

Towards analysis and refinement of auto-tuning spaces

Autoři

FILIPOVIČ, Jiří; Suren Harutyunyan GEVORGYAN; Eduardo CESAR a Anna SIKORA

Vydání

PARALLEL COMPUTING, AMSTERDAM, ELSEVIER, 2026, 0167-8191

Další údaje

Jazyk

angličtina

Typ výsledku

Článek v odborném periodiku

Obor

10201 Computer sciences, information science, bioinformatics

Stát vydavatele

Nizozemské království

Utajení

není předmětem státního či obchodního tajemství

Odkazy

URL

Impakt faktor

Impact factor: 2.100 v roce 2024

Označené pro přenos do RIV

Ano

Organizační jednotka

Ústav výpočetní techniky

DOI

https://doi.org/10.1016/j.parco.2026.103185

UT WoS

001674763400001

Klíčová slova anglicky

Auto-tuning; Tuning spaces; Hardware performance counters

Štítky

rivok

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 25. 3. 2026 11:54, Mgr. Eva Špillingová

Anotace

V originále

Source code-level auto-tuning enables applications to adapt their implementation to maintain peak performance under varying execution environments (i. e.hardware, input, or application settings). However, the performance of the auto-tuned code is inherently tied to the design of the tuning space (the space of possible changes to the code). An ideal tuning space must include configurations diverse enough to ensure high performance across all targeted environments while simultaneously eliminating redundant or inefficient regions that slow the tuning space search process. Traditional research has focused primarily on identifying optimization opportunities in the code and on efficient tuning space search. However, there is no rigorous methodology or tool supporting analysis and refinement of the tuning spaces, allowing for the addition of configurations that perform well in an unseen environment or the removal of configurations that perform poorly in any realistic environment. In this short communication, we argue that hardware performance counters should be used to analyze tuning spaces, and that such an analysis would allow programmers to refine the tuning spaces by adding configurations that unlock additional performance in unseen environments and removing those unlikely to produce efficient code in any realistic environment. While our primary goal is to introduce this research question and foster discussion, we also present a preliminary methodology for tuning-space analysis. We validate our approach through a case study using a GPU implementation of an N-body simulation. Our results demonstrate that the proposed analysis can detect the weaknesses of a tuning space: based on its outcomes, we refined the tuning space, improving the average configuration performance 3.3x, and the best-performing configuration by 2-18%.

Návaznosti

LM2023054, projekt VaV

Název: e-Infrastruktura CZ

Investor: Ministerstvo školství, mládeže a tělovýchovy ČR, e-Infrastruktura CZ

Citovat

FILIPOVIČ, Jiří; Suren Harutyunyan GEVORGYAN; Eduardo CESAR a Anna SIKORA. Towards analysis and refinement of auto-tuning spaces. PARALLEL COMPUTING. AMSTERDAM: ELSEVIER, 2026, roč. 127, March, s. 1-6. ISSN 0167-8191. Dostupné z: https://doi.org/10.1016/j.parco.2026.103185.

@article{2554359,
   author = {Filipovič, Jiří and Gevorgyan, Suren Harutyunyan and Cesar, Eduardo and Sikora, Anna},
   article_location = {AMSTERDAM},
   article_number = {March},
   doi = {https://doi.org/10.1016/j.parco.2026.103185},
   keywords = {Auto-tuning; Tuning spaces; Hardware performance counters},
   language = {eng},
   issn = {0167-8191},
   journal = {PARALLEL COMPUTING},
   title = {Towards analysis and refinement of auto-tuning spaces},
   url = {https://www.sciencedirect.com/science/article/pii/S0167819126000037?casa_token=OOcbgu2QT2IAAAAA:yxWVJJnU9e8GIwGbNogFdrZPu9w0kPLRGVUavVf6VESC3y_WIy1HjbVt7BvUxqO709vOONSE},
   volume = {127},
   year = {2026}
}

TY  - JOUR
ID  - 2554359
AU  - Filipovič, Jiří - Gevorgyan, Suren Harutyunyan - Cesar, Eduardo - Sikora, Anna
PY  - 2026
TI  - Towards analysis and refinement of auto-tuning spaces
JF  - PARALLEL COMPUTING
VL  - 127
IS  - March
SP  - 1-6
EP  - 1-6
PB  - ELSEVIER
SN  - 01678191
KW  - Auto-tuning
KW  - Tuning spaces
KW  - Hardware performance counters
UR  - https://www.sciencedirect.com/science/article/pii/S0167819126000037?casa_token=OOcbgu2QT2IAAAAA:yxWVJJnU9e8GIwGbNogFdrZPu9w0kPLRGVUavVf6VESC3y_WIy1HjbVt7BvUxqO709vOONSE
N2  - Source code-level auto-tuning enables applications to adapt their implementation to maintain peak performance under varying execution environments (i. e.hardware, input, or application settings). However, the performance of the auto-tuned code is inherently tied to the design of the tuning space (the space of possible changes to the code). An ideal tuning space must include configurations diverse enough to ensure high performance across all targeted environments while simultaneously eliminating redundant or inefficient regions that slow the tuning space search process. Traditional research has focused primarily on identifying optimization opportunities in the code and on efficient tuning space search. However, there is no rigorous methodology or tool supporting analysis and refinement of the tuning spaces, allowing for the addition of configurations that perform well in an unseen environment or the removal of configurations that perform poorly in any realistic environment. In this short communication, we argue that hardware performance counters should be used to analyze tuning spaces, and that such an analysis would allow programmers to refine the tuning spaces by adding configurations that unlock additional performance in unseen environments and removing those unlikely to produce efficient code in any realistic environment. While our primary goal is to introduce this research question and foster discussion, we also present a preliminary methodology for tuning-space analysis. We validate our approach through a case study using a GPU implementation of an N-body simulation. Our results demonstrate that the proposed analysis can detect the weaknesses of a tuning space: based on its outcomes, we refined the tuning space, improving the average configuration performance 3.3x, and the best-performing configuration by 2-18%.
ER  -

FILIPOVIČ, Jiří; Suren Harutyunyan GEVORGYAN; Eduardo CESAR a Anna SIKORA. Towards analysis and refinement of auto-tuning spaces. \textit{PARALLEL COMPUTING}. AMSTERDAM: ELSEVIER, 2026, roč.~127, March, s.~1-6. ISSN~0167-8191. Dostupné z: https://doi.org/10.1016/j.parco.2026.103185.

Přehled o publikaci