Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.08372

vTen: Tensor-Centric Verification Framework for Domain-Specific Accelerators

Chanmin Baek·Keehyuk Lee·Mincheol Cha·Somi Hong·Xuan Truong Nguyen·Hyuk-Jae Lee
View original source ↗Read source document ↗Export metadata ↓

Abstract

The semantic gap between tensor-centric software models and signal-level hardware testbenches creates significant productivity bottlenecks in verifying Domain-Specific Accelerators (DSAs). Existing frameworks like Cocotb suffer from prohibitive synchronization overheads due to fine-grained interactions. To address this, we propose vTen, a data-centric framework that strictly decouples verification intent from execution mechanics. By leveraging a declarative DSL and kernel-granular batching, vTen minimizes host-simulator interaction frequency. Evaluation on a production-scale 3D U-Net accelerator demonstrates that vTen achieves a 2x performance improvement in simulation latency and a 60.3% reduction in code complexity compared to Cocotb.

Explore related subjects

cs.DCcs.AR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Chanmin Baek, Keehyuk Lee, Mincheol Cha, Somi Hong, Xuan Truong Nguyen, Hyuk-Jae Lee. 2026-10-06. vTen: Tensor-Centric Verification Framework for Domain-Specific Accelerators. https://arxiv.org/abs/2610.08372

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving

Mooncake is the serving platform for Kimi, a leading LLM service provided by Moonshot AI. It features a KVCache-centric disaggregated architecture that separates the prefill and decoding clusters. It also leverages the underutilized CPU, DRAM, and SSD resources of the GPU cluster to implement a disaggregated cache of KVCache. The core of Mooncake is its KVCache-centric scheduler, which balances maximizing overall effective throughput while meeting latency-related Service Level Objectives (SLOs). Unlike traditional studies that assume all requests will be processed, Mooncake faces challenges due to highly overloaded scenarios. To mitigate these, we developed a prediction-based early rejection policy. Experiments show that Mooncake excels in long-context scenarios. Compared to the baseline method, Mooncake can achieve up to a 525% increase in throughput in certain simulated scenarios while adhering to SLOs. Under real workloads, Mooncake's innovative architecture enables Kimi to handle 75% more requests.

cs.DC↗
arXiv2026

TAPAAL SMC: Statistical Model Checking of Stochastic Timed-Arc Petri Nets

Timed-Arc Petri net (TAPN) is a timed extension of the classical Petri net model where tokens have their age and input arcs are associated with time intervals restricting the ages of tokens available for transition firing. Additionally, a TAPN can also contain place invariants constraining the ages of tokens in places, inhibitor arcs preventing a transition from firing and transport arcs that preserve token ages upon firing. This set of features, as much as it allows us to model complex systems, also often makes verification problems computationally hard or even undecidable. Moreover, in order to model real-life examples, additional stochastic aspects are often necessary to capture the desired behaviour. We suggest the first stochastic semantics for TAPNs and design and implement the quantitative and qualitative Statistical Model Checking (SMC) algorithms in the model checker TAPAAL. We argue for the semantic choices we made in the stochastic semantics and prove that the semantics is well-behaving. On a number of case studies we demonstrate the practical applicability of our modelling formalism and its SMC implementation.

cs.DC↗
arXiv2026

eAVID: Asynchronous Verifiable Information Dispersal with Post-Dissemination Pruning

The well-known Asynchronous Verifiable Information Dispersal (AVID) problem lets a sender disperse a message across $N=3F+1$ nodes such that it remains recoverable despite up to $F$ Byzantine failures and unbounded message delays. An optimal AVID scheme requires $3\times$ the message size in communication and storage: up to $F$ nodes may be delayed in responding, and among the remaining $2F+1$ respondents, up to $F$ may be Byzantine. This paper presents eAVID, an elastic AVID scheme that provides two key improvements. First, after dissemination, nodes can prune up to half of their stored information without requiring anyone to reconstruct or recode the message. Second, pruning is enabled by an extremely simple protocol. Nodes collect acknowledgments from one another confirming receipt of their coded pieces, after which each node locally prunes its stored information. eAVID achieves this $2\times$ reduction while making two tradeoffs: the sender generates $2N$ fragments rather than $N$, and pruning may necessitate contacting $2F+1$ nodes for message retrieval rather than $F+1$. eAVID was implemented in DispersedSimplex, a simple-to-understand BFT consensus protocol that erasure-codes its blocks. The implementation demonstrates two features. First, eAVID achieves these improvements using a flat erasure-coding scheme that requires no metadata or bookkeeping at the nodes during reconstruction. Second, it delivers the storage savings without a performance cost: with all nodes responsive, pruning fires on over $99\%$ of blocks and steady-state per-node storage falls by $42$-$46\%$ for committees of $10$ to $22$ nodes. Throughput and latency match the unmodified protocol showing that we can achieve these storage savings without any performance costs.

cs.DC↗