arXiv subjects

Barry Lyu

Publications and source records attributed to Barry Lyu.

5 recordsLinked to original sources

arXiv2026

Fengshui: Demystifying Chiplet Ecosystem and Bespoke Neural Network Accelerator Codesign

Modern ML workloads, with stringent latency and energy constraints, are increasingly hard to run efficiently on homogeneous commodity hardware. We argue that operator-level disaggregation--tailoring microarchitecture, batching, and memory hierarchy to each operator--is essential to overcome these limitations, though the resulting highly bespoke accelerators incur prohibitive Non-Recurring Engineering (NRE) costs. Chiplet-based integration amortizes NRE across applications, but choosing which chiplets to build and how to compose them into accelerators is circularly dependent--a chiplet pool's value depends on the constructed accelerators, while accelerator quality is constrained by available chiplets. This paper introduces Fengshui, a chiplet ecosystem and accelerator co-design framework that jointly optimizes chiplet pool composition and bespoke application-specific integrated circuit (BASIC) design. Fengshui constructs BASICs through operator-level disaggregation, co-exploring chiplet and memory heterogeneity, tensor fusion, and pipeline/tensor/expert parallelism with place-and-route validation for physical implementability. With just 8 strategically selected chiplets, encompassing network switches, processing-in-memory units, and accelerators with diverse microarchitectures, Fengshui-generated BASICs achieve 48.5%, 88.1%, 93.0%, and 97.8% reductions in energy, energy-cost product (EC), energy-delay product (EDP), and energy-delay-cost product (EDPC) over homogeneous accelerators, while scoring within 4.1% of unconstrained heterogeneous designs across diverse neural networks. For datacenter MoE and dense LLM serving, Fengshui reduces prefill energy and EC by up to 16.8% and 28.7%, respectively; for edge autonomous vehicle perception, it achieves 12.0% energy and 23.6% EC reductions under real-time latency constraints.

cs.AR↗

arXiv2026

Rethinking Agentic Kernel Generation for Emerging Accelerators

Emerging accelerators often lack mature compiler backends, motivating neural agents that generate and repair kernels from architectural documentation and simulator feedback. This approach repeatedly reconstructs workload-invariant machine semantics--including instruction behavior, legality constraints, synchronization rules, and memory protocols--for every workload. We argue that these semantics should be compiled once into a persistent symbolic artifact, while neural reasoning should focus on workload-dependent mapping decisions. We present Zomboss, a compiler-mediated agentic framework for kernel generation that places neural search within a verified compiler boundary. Zomboss compiles machine semantics and legality constraints into a reusable mapping interface, then uses a neural agent to optimize workload-dependent decisions within the validated mapping space. Across 20 Gemmini and 36 PLENA workload instances, Zomboss returns a correct verified kernel on all 56 instances. Relative to the compiler default, Zomboss achieves geometric-mean speedups of $3.34\times$ on Gemmini and $1.10\times$ on PLENA. Relative to direct agentic generation, it reduces inference tokens by 71.2% on Gemmini and 54.2% on PLENA. These results show that a compiler-defined symbolic interface turns native kernel synthesis into verified design-space exploration: compiler infrastructure preserves legality and correctness, while neural guidance improves workload-specific performance with lower search cost and complete coverage.

cs.AR↗

arXiv2026

Compute System Organization for High Frequency High Order Wavefront Sensing and Control

Maintaining long-term wavefront stability is critical for the Habitable Worlds Observatory (HWO), which targets contrasts approaching $10^{-10}$ and therefore requires continuous dark-zone maintenance using high-order wavefront sensing and control (HOWFSC). Prior work has advanced HOWFSC algorithms and profiled candidate implementations on radiation-hardened processors, highlighting a substantial gap between the computational demands of LUVOIR-scale HOWFSC and the capabilities of current onboard spacecraft hardware. In this paper, we argue that this gap can be closed by offloading the HOWFSC pipeline to a dedicated co-flying compute satellite at Sun-Earth L2. This approach enables the use of modern, radiation-tolerant high-performance processors without increasing risk to the primary observatory. We show that such an architecture can increase the end-to-end control cadence from the sub-hertz regime typical of radiation-hardened onboard processing or ground-in-the-loop operation to tens and even hundreds of hertz. We evaluate commercial hardware platforms in terms of performance and feasibility, and we propose custom architectures that enable higher control frequencies with significant power consumption reductions. Finally, we outline system-level considerations for co-flying compute, including reliability, satellite integration, and inter-satellite communication constraints.

astro-ph.IM↗

arXiv2026

Opus: Photonic Rail-Optimized Fabric in ML Datacenters

Rail-optimized network fabrics have become the de facto datacenter scale-out fabric for large-scale ML training. However, the use of high-radix electrical switches to provide all-to-all connectivity in rails imposes substantial power and cost. We propose a rethinking of the rail abstraction by retaining its communication semantics, but realizing it using optical circuit switches. The key challenge is that optical switches support one-to-one connectivity at a time, limiting the fan-out of traffic in ML workloads using hybrid parallelisms. We overcome this through \emph{parallelism-driven rail reconfiguration}, which exploits the non-overlapping communication phases of different parallelism dimensions. This time-multiplexes a single set of physical ports across circuit configurations tailored to each phase within a training iteration. We design and implement Opus, a control plane that orchestrates this in-job reconfiguration of photonic rails at parallelism phase boundaries, and evaluate it on a physical OCS testbed, the Perlmutter supercomputer, and in simulation at up to 2,048 GPUs. Our results show that photonic rails can achieve over $23\times$ network power reduction and $4\times$ cost savings while incurring only modest training overhead at production-relevant OCS reconfiguration latencies.

cs.NI↗

arXiv2025

Mozart: A Chiplet Ecosystem-Accelerator Codesign Framework for Composable Bespoke Application Specific Integrated Circuits

Modern AI acceleration faces a fundamental challenge: conventional assumptions about memory requirements, batching effectiveness, and latency-throughput tradeoffs are systemwide generalizations that ignore the heterogeneous computational patterns of individual neural network operators. However, going towards network-level customization and operator-level heterogeneity incur substantial Non-Recurring Engineering (NRE) costs. While chiplet-based approaches have been proposed to amortize NRE costs, reuse opportunities remain limited without carefully identifying which chiplets are truly necessary. This paper introduces Mozart, a chiplet ecosystem and accelerator codesign framework that systematically constructs low cost bespoke application-specific integrated circuits (BASICs). BASICs leverage operator-level disaggregation to explore chiplet and memory heterogeneity, tensor fusion, and tensor parallelism, with place-and-route validation ensuring physical implementability. The framework also enables constraint-aware system-level optimization across deployment contexts ranging from datacenter inference serving to edge computing in autonomous vehicles. The evaluation confirms that with just 8 strategically selected chiplets, Mozart-generated composite BASICs achieve 43.5%, 25.4%, 67.7%, and 78.8% reductions in energy, energy-cost product, energy-delay product (EDP), and energy-delay-cost product compared to traditional homogeneous accelerators. For datacenter LLM serving, Mozart achieves 15-19% energy reduction and 35-39% energy-cost improvement. In speculative decoding, Mozart delivers throughput improvements of 24.6-58.6% while reducing energy consumption by 38.6-45.6%. For autonomous vehicle perception, Mozart reduces energy-cost by 25.54% and energy by 10.53% under real-time constraints.

cs.AR↗