arXiv subjects

Aaron Barnes

Publications and source records attributed to Aaron Barnes.

4 recordsLinked to original sources

arXiv2026

Architecting the Next Generation of Asynchronous, Distributed GPUs for the AI Era

The rapid evolution of machine learning workloads has fundamentally transformed GPU hardware, driving architectures toward Multi-Chip Module (MCM) topologies, asynchronous execution primitives, and persistent, multi-phase kernel behaviors. Despite these shifts, cycle-level simulation infrastructure has lagged behind, lacking the native capability to model the physical non-uniformity of modern GPUs alongside the massive scale of state-of-the-art AI workloads. To bridge this gap, we present a cycle-level simulation framework designed to accurately model modern GPU generations, including Ampere, Hopper, and Blackwell. Rigorously validated against physical silicon, the simulator achieves a 99% Pearson correlation coefficient and a 13.4% mean absolute cycle error on the H100 GPU. Utilizing this infrastructure, we conduct architectural case studies to evaluate emerging design trajectories, including chiplet topology scaling, expanded SRAM capacity and bandwidth, and inter-GPU prefetching strategies.

cs.AR↗

arXiv2026

Triton for MTIA: Bridging the Programming Model Gaps for Custom AI Accelerators

The rapid growth in machine learning workloads has fueled the proliferation of custom accelerator architectures. Designed from the ground up, these accelerators often expose programming models that are distinct from GPUs. While hyperscalers and AI chip startups continue to innovate in this space, achieving broad operator coverage to support diverse models remains a major challenge. Additionally, an easy-to-use, high-level kernel programming language is important for rapid iteration of models and kernels. Triton, together with TorchInductor, addresses these issues on GPUs, but its viability on accelerators with different programming models has yet to be established. In this work, we present the first production-scale application of Triton on a custom ML accelerator, MTIA-2i, developed by Meta. To support MTIA-2i, we develop a new compiler backend that targets it, introduce enhancements to TorchInductor code generation, and propose minimal language extensions that expose MTIA-specific architectural features. We demonstrate that Triton-MTIA kernels achieve performance competitive with expert-tuned C++ implementations. Leveraging these development efficiency gains, we successfully deployed manually written and Inductor-generated Triton kernels in production across approximately 60 different model types, accounting for 50% of layers and 47% of non-GEMM execution time for these models. Our results provide compelling evidence that DSLs like Triton can bridge the programming model gaps between ML frameworks, kernels, and custom accelerators, enabling rapid innovation and efficient deployment at scale.

cs.PL↗

arXiv2024

RayFlex: An Open-Source RTL Implementation of the Hardware Ray Tracer Datapath

The advent of hardware ray tracing (RT) units has brought unprecedented realism to real-time rendered computer graphics. However, the potential of these units extends beyond graphics, offering acceleration for various computational tasks such as tree traversal and nearest-neighbor search. We introduce RayFlex, a first-of-its-kind open-source RTL implementation of a hardware ray tracer datapath designed to facilitate research in general-purpose programmable RT units. RayFlex's architecture is both extensible and flexible, thanks to two core design concepts: the parameterized RayFlex Skid Buffer module and the "defined-once-instantiated-everywhere" Shared RayFlex Data Structure. This makes RayFlex an ideal testing ground for academic research and exploration. Our implementation allows researchers to explore various design choices, fostering a realistic understanding of hardware ray tracer design trade-offs. Through comprehensive case studies, we demonstrate the versatility of RayFlex in evaluating different pipeline configurations and extending its functionality to support additional computational tasks. We show that by extending the functionality of a baseline RT unit datapath with an area cost of 36 percent and a power overhead of about 20 percent, the RT unit can calculate the Euclidean distance and cosine distance of vectors of arbitrary dimension, thereby accelerating a broader range of data-analytics workloads. The source code of RayFlex is available at https://github.com/purdue-aalp/rayflex.

cs.AR↗

arXiv2011

Thermally Activated Magnetization and Resistance Decay during Near Ambient Temperature Aging of Co Nanoflakes in a Confining Semi-metallic Environment

We report the observation of magnetic and resistive aging in a self assembled nanoparticle system produced in a multilayer Co/Sb sandwich. The aging decays are characterized by an initial slow decay followed by a more rapid decay in both the magnetization and resistance. The decays are large accounting for almost 70% of the magnetization and almost 40% of the resistance for samples deposited at 35 $^oC$. For samples deposited at 50 $^oC$ the magnetization decay accounts for $\sim 50%$ of the magnetization and 50% of the resistance. During the more rapid part of the decay, the concavity of the slope of the decay changes sign and this inflection point can be used to provide a characteristic time. The characteristic time is strongly and systematically temperature dependent, ranging from $\sim1$x$10^2 s$ at 400K to $\sim3$x$10^5 s$ at 320K in samples deposited at $35 ^oC$. Samples deposited at 50 $^oC$ displayed a 7-8 fold increase in the characteristic time (compared to the $35 ^oC$ samples) for a given aging temperature, indicating that this timescale may be tunable. Both the temperature scale and time scales are in potentially useful regimes. Pre-Aging, Scanning Tunneling Microscopy (STM) reveals that the Co forms in nanoscale flakes. During aging the nanoflakes melt and migrate into each other in an anisotropic fashion forming elongated Co nanowires. This aging behavior occurs within a confined environment of the enveloping Sb layers. The relationship between the characteristic time and aging temperature fits an Arrhenius law indicating activated dynamics.

cond-mat.mes-hall↗