An official website of the United States government
Not AvailableModern Artificial Intelligence (AI) workloads demand computing systems with large silicon area to sustain throughput and competitive performance. However, prohibitive manufacturing costs and yield limitations at advanced tech nodes and die-size reaching the reticle limit restrain us from achieving this. With the recent innovations in advanced packaging technologies, chiplet-based architectures have gained significant attention in the AI hardware domain. However, the vast design space of chiplet-based AI accelerator design and the absence of system and package-level co-design methodology make it difficult for the designer to find the optimum design point regarding Power, Performance, Area, and manufacturing Cost (PPAC). This paper presents Chiplet-Gym, a Reinforcement Learning (RL)-based optimization framework to explore the vast design space of chiplet-based AI accelerators, encompassing the resource allocation, placement, and packaging architecture. We analytically model the PPAC of the chiplet-based AI accelerator and integrate it into an OpenAI gym environment to evaluate the design points. We also explore non-RL-based optimization approaches and combine these two approaches to ensure the robustness of the optimizer. The optimizer-suggested design point achieves 1.52× throughput, 0.27× energy, and 0.89× cost of its monolithic counterpart at iso-area.
more »
« less
System and Design Technology Co-optimization of Chiplet-based AI Accelerator with Machine Learning
With the availability of advanced packaging technology and its attractive features, the chiplet-based architecture has gained traction among chip designers. The large design space and the lack of system and package-level co-design methods make it difficult for the designers to create the optimum design choices. In this research, considering the colossal design space of advanced packaging technologies, resource allocation, and chiplet placement, we design an optimizer that looks for the design choices that maximize the Power, Performance, and Area (PPA) and minimize the cost of the chiplet-based AI accelerator. Inspired by the Bayesian approach for black-box function optimization, our optimizer guides the search space toward global maxima instead of randomly traversing through the search space. We analytically synthesize a dataset from the search space and train an ML model to predict the target value of our defined cost function at the optimizer-suggested points. The optimizer locates the optimum design choices from the specified search space (≥ 1M data points) with minimal iterations (≤ 200 iterations) and trivial run time.
more »
« less
- Award ID(s):
- 2153394
- PAR ID:
- 10428292
- Date Published:
- Journal Name:
- Proceedings of the Great Lakes Symposium on VLSI
- Page Range / eLocation ID:
- 697 to 702
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Parametric optimization techniques allow building designers to pursue multiple performance objectives, which can benefit the overall design. However, the strategies used by architecture and engineering graduate students when working with optimization tools are unclear, and ineffective computational design procedures may limit their success as future designers. In response, this re-search identifies several designerly behaviors of graduate students when responding to a conceptual building design optimization task. It uses eye-tracking, screen recording, and empirical methods to code their behaviors following the situated FBS framework. From these data streams, three different types of design iterations emerge: one by the designer alone, one by the optimizer alone, and one by the designer incorporating feedback from the optimizer. Based on the timing and frequency of these loops, student participants were characterized as completing partial, crude, or complete optimization cycles while developing their designs. This organization of optimization techniques establishes reoccurring strategies employed by developing designers, which can encourage future pedagogical approaches that empower students to incorporate complete optimization cycles while improving their designs. It can also be used in future research studies to establish clear links between types of design optimization behavior and design quality.more » « less
-
Fast-evolving artificial intelligence (AI) algorithms such as large language models have been driving the ever increasing computing demands in today’s data centers. Heterogeneous computing with domain-specific architectures (DSAs) brings many opportunities when scaling up and scaling out the computing system. In particular, heterogeneous chiplet architecture is favored to keep scaling up and scaling out the system as well as to reduce the design complexity and the cost stemming from the traditional monolithic chip design. However, how to interconnect computing resources and orchestrate heterogeneous chiplets is the key to success. In this paper, we first discuss the diversity and evolving demands of different AI workloads. We discuss how chiplet brings better cost efficiency and shorter time to market. Then we discuss the challenges in establishing chiplet interface standards, packaging, and security issues. We further discuss the software programming challenges in chiplet systems.more » « less
-
Building facades are components that shape a structure’s daylighting, energy use, and view factors. This paper presents an approach that enables designers to understand the impact that different facade designs will have over time and space in the built environment through a BIM-enabled augmented reality system. The system permits the examination of a range of facade retrofit scenarios and visualizes the daylighting simulations and aesthetics of a structure while retaining function and comfort. A focus of our study was to measure how participants make decisions within the multiobjective decision space designers often face when buildings undergo retrofitting. This process often requires designers to search for a set of alternatives that represent the optimal solution. We analyze the decision-making process of forty-four subjects to determine how they explore design choices. Our results indicate the feasibility of using BIM-enabled AR to improve how designers make informed decisions.more » « less
-
null (Ed.)Abstract Designers make information acquisition decisions, such as where to search and when to stop the search. Such decisions are typically made sequentially, such that at every search step designers gain information by learning about the design space. However, when designers begin acquiring information, their decisions are primarily based on their prior knowledge. Prior knowledge influences the initial set of assumptions that designers use to learn about the design space. These assumptions are collectively termed as inductive biases. Identifying such biases can help us better understand how designers use their prior knowledge to solve problems in the light of uncertainty. Thus, in this study, we identify inductive biases in humans in sequential information acquisition tasks. To do so, we analyze experimental data from a set of behavioral experiments conducted in the past [1–5]. All of these experiments were designed to study various factors that influence sequential information acquisition behaviors. Across these studies, we identify similar decision making behaviors in the participants in their very first decision to “choose x”. We find that their choices of “x” are not uniformly distributed in the design space. Since such experiments are abstractions of real design scenarios, it implies that further contextualization of such experiments would only increase the influence of these biases. Thus, we highlight the need to study the influence of such biases to better understand designer behaviors. We conclude that in the context of Bayesian modeling of designers’ behaviors, utilizing the identified inductive biases would enable us to better model designer’s priors for design search contexts as compared to using non-informative priors.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1145/3583781.3590233
