Nov 2026· Anais do XXVII Simpósio em Sistemas Computacionais de Alto Desempenho (SSCAD 2026)· pp. 455-468· 0 citations· 26 references
TL;DR
A heuristic that reduces the design space via an energy-accuracy trade-off, combined with machine learning-based models that replace simulation during DSE is proposed, demonstrating that instruction-level approximate DSE can be made tractable without meaningful loss of solution quality.
Abstract
Approximate Computing (AC) trades reduced precision for energy savings in error-tolerant scenarios, but identifying the right instruction-level placement demands extensive, computationally prohibitive Design Space Exploration (DSE), further burdened by costly simulation-based evaluation. To address this, we propose a heuristic that reduces the design space via an energy-accuracy trade-off, combined with machine learning-based models that replace simulation during DSE. We introduced custom analysis passes to extract program features for these models, and trained predictors achieved R2 ≥ 0.955 for error estimation and R2 ≥ 0.986 for energy estimation across all benchmarks, with Spearman rank correlations at least ρ = 0.966, confirming that solution ordering is faithfully preserved against full simulation. Replacing simulation with ML-based prediction yielded mean speedups from ≈ 23,987× to ≈ 823,693×, demonstrating that instruction-level approximate DSE can be made tractable without meaningful loss of solution quality.
NPU design-space exploration for LLM inference mainly relies on an expensive simulator that often returns only scalar feedback such as latency and power. Such feedback can rank evaluated configurations, but gives little information for deciding why a configuration is slow or which architecture parameter should be chang...
Yi-Cheng He, Yu Xue, Tian-Yang Wu et al.· 0 citations
Virtual memory research increasingly relies on empirical performance models that use simulated address translation metrics (x)---such as TLB miss rate and page-walk cycles---to predict benchmark runtimes (y) for new hardware designs. The accuracy of these models depends on the training data: a set of (x,y) pairs collec...
Mohammad Agbarya, Idan Yaniv, Dan Tsafrir· Proceedings of the 32nd ACM...· 0 citations
When optimizing computing architecture for specific computationally intensive tasks, such as evaluating or training a neural network, it is important to identify what computational subtasks will offer the greatest decrease in cost (e.g., wall clock time or energy usage). This problem is known as Hardware-Software (HS)...
Cameron Ibrahim, S. Ferdous, Erdal Mutlu et al.· 0 citations
Test-time scaling has recently emerged as a powerful approach for improving LLM reasoning by allocating additional computation during inference, substantially enhancing accuracy on challenging tasks such as mathematics and coding. To accelerate the exploration of reasoning paths, recent studies proposed speculative exe...
As Moore's law reaches its physical and economic limits, domain-specific approaches are increasingly employed to accelerate machine learning workloads. Hyperdimensional Computing (HDC) represents one such emerging paradigm, offering an alternative to conventional deep learning techniques. Rooted in cognitive models of...
Xavier Routh, Abdul Rafae Noor, Akash Kothari et al.· Proceedings of the 32nd ACM...· 0 citations
Large Language Models (LLMs) are increasingly used to optimize code for performance. This practice often assumes that improvements in execution speed naturally translate into overall lower energy consumption. We challenge this assumption empirically by measuring whether performance-optimized code produced by LLM self-r...
Sadra Emami, Benjamin Weigell, Mohammad Reza Mousavi et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.