An LLM-agent pipeline with profiling, binary analysis, and SMT simulation automatically parallelizes latency-critical benchmarks via the Relic framework, reporting a 17% geomean gain after excluding failures.
Speculative precomputation: long -range prefetching of delinquent loads,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.DC 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Accelerating Latency-Critical Applications with AI-Powered Semi-Automatic Fine-Grained Parallelization on SMT Processors
An LLM-agent pipeline with profiling, binary analysis, and SMT simulation automatically parallelizes latency-critical benchmarks via the Relic framework, reporting a 17% geomean gain after excluding failures.