Searcharxiv⌕ Search

arXiv subjects

Yimin Gao

Publications and source records attributed to Yimin Gao.

9 recordsLinked to original sources

Mutable Inference: Architecting LLM Accelerators for Inference-Time Learning

Long-context LLMs must retain growing histories, but full attention makes each new token increasingly expensive. Inference-time learning (ITL) offers another approach. The model keeps training on the context during inference, retaining what it learns in its weights. Recent work shows that ITL can match full-context attention quality while attending only to a recent fixed-size window. This bounds attention costs but adds Learning (backward computation and weight updates) alongside Prediction (the forward pass) costs. Inference accelerators assume fixed weights, while training accelerators target complete training steps. This work identifies two opportunities. Attention's non-matrix operations leave enough idle matrix capacity for Learning, while selected layers can tolerate delayed weight updates with little quality loss. We exploit both with Learning While Serving (LWS), an inference accelerator architecture that shares compute hardware between latency-critical Prediction and deadline-flexible Learning. Hardware-managed per-layer weight versioning allows Learning to publish updates while in-flight computations read older versions. Resumable reduction blocks let Learning kernels span fragmented idle windows and yield to Prediction, while dedicated softmax-gradient units relieve vector contention. On modeled datacenter and edge NPUs, LWS hides Learning within Prediction, sustaining Prediction-only throughput while preserving 99.3% of fully fresh Learning's quality benefit for TTT-E2E 3B at 128K tokens. Across nine models, LWS achieves 1.6x the throughput of serial execution and 1.3x that of the best prior co-execution architecture. RTL synthesis shows that LWS adds 1.5% to NPU core area without another MAC array. By efficiently turning idle inference capacity into learning capacity, LWS establishes inference-time learning as a first-class workload for future LLM accelerators.

cs.AR↗

SLED-IFV: Solver-Validated LLM-Guided Decomposition for Scalable Hardware Information-Flow Verification

Formal hardware information-flow verification (IFV) provides strong guarantees against secret-dependent timing and control behavior, but often scales poorly on realistic RTL. We identify two recurring proof barriers in self-composed IFV: implementation complexity, where proof-hard datapath logic dominates even though the property needs only a compact boundary relation, and relational inductive complexity, where the proof depends on cross-copy public-control facts that the backend prover does not infer efficiently. To address them, we introduce two semantic proof decomposition forms: functional simplification, which replaces a proof-hard RTL region with a validated over-approximate summary, and relational strengthening, which exposes and proves the cross-copy relations needed for induction. We further present SLED-IFV, a solver-validated LLM-guided flow that automates the selection of these forms and their concrete targets. Given a self-composed miter and an oracle-free decision sheet, the LLM proposes a decomposition, then materializes it into proof artifacts under controller checks. The controller compiles the checked artifacts into proof obligations, and the formal verification backend remains the sole authority for acceptance. Across nine nontrivial benchmarks constructed from real RTL, SLED-IFV achieves up to 603x solver-only speedup and converts two 12-hour timeouts into completed proofs. The closed-loop flow produces verifier-accepted decompositions for all cases.

cs.CR↗

FlexPosit: Tunable Fractional Precision for LLM Inference Accelerators

Large language models (LLMs) offer remarkable capabilities but impose prohibitive compute and energy costs. Quantization governs the trade-offs between accuracy and hardware efficiency across granularity and bit-width. Finer granularity (e.g., group-wise) provides high accuracy but incurs scaling and control overhead, while coarser granularity (e.g., channel-wise) has lower overhead but loses accuracy at low precision. Meanwhile, mixed-precision quantization exposes rich accuracy-efficiency trade-offs algorithmically, but existing LLM accelerators remain limited to discrete precision modes, leaving the fractional design space between them unexplored. FlexPosit bridges these gaps through co-design of Posit-based quantization and a precision-tunable bit-serial architecture. Algorithmically, FlexPosit employs distribution-aware quantization with hardware-aligned, sensitivity-guided mixed-precision allocation, leveraging the Posit format's tapered precision to achieve group-wise-like accuracy with channel-wise-like regularity. Architecturally, FlexPosit is a unified bit-serial systolic array with lightweight per-column decoders, unified Processing Elements (PEs), and a global precision controller, enabling tunable fractional precision while preserving fully regular systolic dataflow. Across diverse LLMs, FlexPosit achieves near-FP16 accuracy with sub-5-bit fractional weights. It achieves 1.8x higher throughput and 1.2x lower energy than BitMoD (group-wise quantization), and 1.5x higher throughput and 2.0x lower energy than OliVe (channel-wise quantization), establishing a new Pareto frontier for precision-tunable LLM acceleration.

cs.AR↗

Modeling and Resource Optimization for Quantum Oracles

Quantum oracles are fundamental building blocks of many quantum algorithms, and their resource consumption directly affects performance, yet structured description and complexity analysis for their composition are still lacking. In this paper, we introduce the Framework for Oracle Recursion Modeling (FORM), a unified formal abstraction of multi-function composition in quantum oracles: it provides a structured description of the composition layer, makes its gate complexity exactly computable, and turns oracle design into an optimizable tree-construction problem. Based on this model, we propose the ShallowGrow algorithm, which constructs an oracle structure under a given ancilla budget and provably minimizes the number of function evaluations. On Boolean quadratic equation systems, ShallowGrow reduces Qiskit-measured circuit depth by 54.1% on average relative to the state-of-the-art W-cycle construction, with consistent reductions on the EPFL and ISCAS85 combinational logic networks under scarce ancilla budgets. Furthermore, pebbling-based syntheses trade space against time within the logic network of a function; ShallowGrow extends this trade-off across functions, and integrated with their published circuits it reduces the ancillary qubits of a complete oracle from one per constraint function to logarithmically many. With half as many ancillas as constraint functions, the T-count falls by a factor of 7.9 to 32 relative to the W-cycle-based construction.

quant-ph↗

High-Performance Low-Power Adiabatic Systolic Array Design in Advanced FinFET Nodes

Adiabatic logic has traditionally been recognized as a low-power solution but constrained to low clock speeds to preserve adiabatic behavior. For advanced FinFET nodes, however, clock frequencies have plateaued due to power/thermal concerns (dark silicon) even as the intrinsic device speeds have continued to scale. This convergence opens an opportunity for adiabatic logic to maintain adiabatic behavior even at GHz clocks. We demonstrate an adiabatic logic (AL) design methodology through a MAC systolic array implemented in commercial 16 nm FinFET technology with a resonant 4-phase power clock (PCK) generator, including digital-to-AL and AL-to-digital interfaces. Simulations show that the AL MAC systolic array at 1 GHz achieves power reductions of up to 42% and 36% at the core and system levels, respectively, compared to digital counterparts. Scaling to more advanced nodes should provide even better power/performance metrics.

cs.AR↗

Guarded Equivalence Predicates for Scalable Formal Hardware Information-Flow Verification

Formal hardware information-flow verification is a principled way to rule out secret-dependent functional or timing observations, but scaling such proofs remains difficult. Self-composition reduces information-flow verification to safety checking over two circuit copies, creating relational proof obligations that are hard for a generic PDR engine to discover from bit-level logic alone. Recent PDR-based techniques exploit this duplicated structure through copy symmetry and global cross-copy equivalence predicates. These predicates are effective when corresponding internal signals agree throughout the reachable state space, but they do not capture equalities that are relevant only in a specific control context. We observe that such contextual relations arise naturally in hardware IFV proofs: an internal signal pair may need to agree only within a control phase, transaction window, loop state, or protocol region. We introduce guarded equivalence predicates to expose these relations to PDR. Rather than treating a proposed contextual equality as an assumption, the verifier submits the corresponding mismatch condition as an auxiliary blocking obligation. Guards are proposed from relational counterexamples-to-induction using CTI-local extraction and state-split search; only candidates proved unreachable by the backend affect the proof. Across 12 IFV benchmarks and two PDR backends, guarded predicates convert two contextual baseline timeouts into completed proofs within 34.2--89.5s under an 1800s limit, while reducing proof time by up to 10.8x on additional benchmarks.

cs.CR↗

EDA-Q: Electronic Design Automation for Superconducting Quantum Chip

Electronic Design Automation (EDA) plays a crucial role in classical chip design and significantly influences the development of quantum chip design. However, traditional EDA tools cannot be directly applied to quantum chip design due to vast differences compared to the classical realm. Several EDA products tailored for quantum chip design currently exist, yet they only cover partial stages of the quantum chip design process instead of offering a fully comprehensive solution. Additionally, they often encounter issues such as limited automation, steep learning curves, challenges in integrating with actual fabrication processes, and difficulties in expanding functionality. To address these issues, we developed a full-stack EDA tool specifically for quantum chip design, called EDA-Q. The design workflow incorporates functionalities present in existing quantum EDA tools while supplementing critical design stages such as device mapping and fabrication process mapping, which users expect. EDA-Q utilizes a unique architecture to achieve exceptional scalability and flexibility. The integrated design mode guarantees algorithm compatibility with different chip components, while employing a specialized interactive processing mode to offer users a straightforward and adaptable command interface. Application examples demonstrate that EDA-Q significantly reduces chip design cycles, enhances automation levels, and decreases the time required for manual intervention. Multiple rounds of testing on the designed chip have validated the effectiveness of EDA-Q in practical applications.

cs.ET↗

Membrane: Accelerating Database Analytics with Bank-Level DRAM-PIM Filtering

In-memory database query processing frequently involves substantial data transfers between the CPU and memory, leading to inefficiencies due to Von Neumann bottleneck. Processing-in-Memory (PIM) architectures offer a viable solution to alleviate this bottleneck. In our study, we employ a commonly used software approach that streamlines JOIN operations into simpler selection or filtering tasks using pre-join denormalization which makes query processing workload more amenable to PIM acceleration. This research explores DRAM design landscape to evaluate how effectively these filtering tasks can be efficiently executed across DRAM hierarchy and their effect on overall application speedup. We also find that operations such as aggregates are more suitably executed on the CPU rather than PIM. Thus, we propose a cooperative query processing framework that capitalizes on both CPU and PIM strengths, where (i) the DRAM-based PIM block, with its massive parallelism, supports scan operations while (ii) CPU, with its flexible architecture, supports the rest of query execution. This allows us to utilize both PIM and CPU where appropriate and prevent dramatic changes to the overall system architecture. With these minimal modifications, our methodology enables us to faithfully perform end-to-end performance evaluations using established analytics benchmarks such as TPCH and star-schema benchmark (SSB). Our findings show that this novel mapping approach improves performance, delivering a 5.92x/6.5x speedup compared to a traditional schema and 3.03-4.05x speedup compared to a denormalized schema with 9-17% memory overhead, depending on the degree of partial denormalization. Further, we provide insights into query selectivity, memory overheads, and software optimizations in the context of PIM-based filtering, which better explain the behavior and performance of these systems across the benchmarks.

cs.AR↗

Changes in porosity, permeability and surface area during rock dissolution: effects of mineralogical heterogeneity

Effects of heterogeneity of mineral distribution and reaction rate on the rock dissolution process are investigated using a pore-scale reactive transport model based on the lattice Boltzmann method. Coupled fluid flow, species transport, chemical reaction and solid structure alternation due to dissolution are simulated. Effects of mineral distributions and chemical heterogeneity on the dissolution behaviors and evolutions of hydrologic properties are studied under different reactive transport conditions. Simulation results show that the coupling between advection, diffusion and reaction as well as the mineralogical heterogeneity leads to complex reactive transport behaviors and complicated temporal evolutions of hydrologic properties including porosity, permeability and reactive surface. Diverse relationships between surface area and volume are predicted, which cannot be described by simple models such as the spherical-grain model. Porosity-permeability relationships also differ under different mineral distributions and reactive transport conditions. Simulation results indicate that it is extremely challenging to propose general relationships for hydrologic properties for dissolution of rocks with mineralogical heterogeneity, due to the complicated interactions between reactive transport and mineralogical heterogeneity.

physics.flu-dyn↗