Huawei Li

Chat-Fusion: An Efficient Batch Inference System for Cloud LLMs Serving

Batch processing is useful for optimizing the utilization of various hardware resources, such as GPUs, CPUs, and other specialized accelerators, and is also important for improving the throughput of LLM serving. Many research works have been conducted for better batching and scheduling policies for the efficiency of cl...

Analysis of LLM Vulnerability to GPU Soft Errors: An Instruction-Level Fault Injection Study

This work conducts the first instruction-level fault injection study of LLM inference to analyze the factors affecting LLM vulnerability and reveals reliability characteristics from multiple perspectives, highlighting the effects of model architecture, parameter scale, and task complexity.

HEARS: High-Fidelity Emulation for Aging Fault via Scan-Chain Reuse and Flip-Flop Sampling

As semiconductor nodes shrink to 7nm and below, aging-induced delay faults (e.g., NBTI, HCI) increasingly threaten SoC reliability. To optimize resilience, designers require early-stage evaluation of fault impacts; however, traditional Fault Injection (FI) assessments are either bottlenecked by the slow execution of pr...

HEAT: A Hybrid Homomorphic Encryption Framework for Secure and Efficient Algorithmic Trading Strategy Hosting

With the rise of algorithmic trading, code-hosting platforms such as TradingView have become increasingly popular for strategy development and deployment. However, these platforms typically store and execute user-submitted strategies in plaintext, which introduces significant risks of data leakage. While fully homomorp...

IGBFuzz: Unified RTL Security Fuzzing via Graph-Aware Bandit Scheduling

E-ALS: Approximation-Potential-Aware E-Graph Rewriting for Approximate Logic Synthesis

E-ALS is proposed, an e-graph-based framework for approximation-aware structural search that combines Function-Reduced Saturation, an ALS-coupled surrogate, search-based extraction, and budget-guided refinement to identify approximation-friendly equivalent structures.

An Efficient Fault-Tolerance Scheme for CKKS Computation on CPUs

This work presents an efficient fault-tolerance scheme for CPU-based CKKS computation that achieves a 100 percent empirical detection rate across 150,000 non-crashing corrupted-result cases and maintains application accuracy close to the fault-free baseline over a wide range of fault rates.

Rtl2lean: Automated RTL-to-Lean Translation with Hierarchical Theorem Generation and Lemma Reuse

Rtl2lean is presented, a framework that automatically translates RTL designs into executable Lean 4 models and builds a hierarchical theorem library for subsequent verification and demonstrates that Rtl2lean can construct machine checked RTL proof libraries with low checking overhead and substantial cross property lemm...