Architecture 2.0 Workshop

About the Workshop

The Architecture 2.0 workshop brings together researchers and practitioners exploring the intersection of agentic artificial intelligence and computer systems. As autonomous agents and AI workflows continue to transform how we design, optimize, and evaluate computing systems, this workshop provides a venue for presenting cutting-edge research and fostering collaboration on agent-based methodologies across the hardware stack.

The workshop focuses on Agentic AI for Systems Design, spanning computer architecture, programming languages, and operating systems. We aim to bring together researchers and practitioners applying agentic tools and autonomous AI techniques to systems challenges, providing a platform for work-in-progress and completed research that advances the state of the art or opens new directions for agent-driven systems engineering.

The workshop centers on a Call for Papers for research submissions, bringing together work that applies agentic AI across architecture and systems design.

📅

Date

Saturday, June 27, 2026

📍

Location

ISCA 2026 · Raleigh, NC
Room 306B

Workshop Schedule

8:00–8:15

Welcome and Opening Remarks

8:15–9:15 Session 1: Agentic EDA — From RTL Optimization to Design-Space Exploration

MAESTRO: A Multi-Agent EDA Orchestrator for Autonomous FPGA Design Closure 8:15–8:30 · Saher Elsayed
Intelligent Equality Saturation: From Hierarchical Prospection to a Minimal Implementation 8:30–8:45 · Youwei Xiao, Chenyun Yin, Yun Liang
AgenticDSE: A Multi-Agent Design Space Exploration Framework with Multi-Phase Bayesian Optimization for Chiplet Accelerators 8:45–9:00 · Zhantong Zhu, Zhuolin Li, Kangbo Bai, Hongou Li, Tianyu Jia
Dr. RTL: Autonomous Agentic RTL Optimization through Tool-Grounded Self-Improvement 9:00–9:15 · Wenji Fang, Zhiyao Xie

9:15–10:00 Session 2: Agentic Systems for Runtime Profiling, Serving, and Diagnosis

Towards Agentic Offline Profiling for Speculative Load Micro-Op Fusion 9:15–9:30 · Deepanjali Mishra
Argus: Agentic, Reference-Calibrated, Tree-Guided, System-Level Bottleneck Localization 9:30–9:45 · Vlad-Petru Nitu, Harsh Songara, Konstantinos Sgouras, Spiros Galanopoulos, Konstantinos Kanellopoulos, Onur Mutlu
Dyserve: Dynamic Strategy Generation for Agent Serving 9:45–10:00 · Jiayi Qian, Zishen Wan, Hanchen Yang, Souvik Kundu, Tushar Krishna

10:00–10:30

Coffee Break (with Poster Session)

10:30–11:15 Plenary Talk: Saman Amarasinghe (MIT)

Talk title: TBD Abstract to be announced.

11:15–12:00 Plenary Talk: Sagar Karandikar (UC Berkeley)

Talk title: TBD Abstract to be announced.

12:00–13:30

Lunch Break

13:30–14:15 Plenary Talk: Dimitrios Skarlatos (CMU)

Talk title: TBD Abstract to be announced.

14:15–15:00 Session 3: Benchmarking and Evaluation of Agentic Architecture

Benchmarking Agentic HLS Design Tasks With HLS-Eval 14:15–14:30 · Stefan Abi-Karam, Callie Hao
Fail2Bench: Turning RTL Agent Failures into a Self-Curating Benchmark 14:30–14:45 · Aryan Chhabra, Sumedh Narahari, Shritan Settipalli, Aadarsh Sivaraman
Punctuated Stagnation 14:45–15:00 · Mahesh Madhav

15:00–15:30

Tutorial — ArchEval: Evaluating AI Agent Capability for Computer Architecture Design

15:30–16:00

Coffee Break (with Poster Session)

16:00–16:45 Plenary Talk: Qijing Jenny Huang (NVIDIA)

Talk title: TBD Abstract to be announced.

16:45–17:30 Session 4: Agentic Optimization for Kernels, Compilers, and Hardware Knowledge

HDLxGraph: Bridging Large Language Models and HDL Repositories via HDL Graph Databases 16:45–17:00 · Jiayin Qin, Pingqing Zheng, Fuqi Zhang, Zishen Wan, Shang Wu, Yu Cao, Caiwen Ding, Yang Katie Zhao
From Skills to Tracelets: Dependency-Guided Transfer for LLM Kernel Agents 17:00–17:15 · Shuoming Zhang, Ruiyuan Xu, Qiuchu Yu, Guangli Li, Huimin Cui, Jiacheng Zhao
AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization 17:15–17:30 · Genghan Zhang, Shaowei Zhu, Anjiang Wei, Zhenyu Song, Allen Nie, Zhen Jia, Nandita Vijaykumar, Yida Wang, Kunle Olukotun

17:30–18:00

Poster Session

Call for Papers: Agentic AI for Architecture & Systems Design

We invite submissions that explore the use of agentic frameworks, autonomous workflows, and artificial intelligence to design, analyze, optimize, or evaluate computer systems. The workshop aims to bring together researchers and practitioners applying agent-based AI techniques to systems challenges across the hardware stack.

We welcome both work-in-progress and completed research that advances the state of the art or opens new directions for agent-driven hardware methodologies and systems research.

Note: The event will not have formal proceedings, and authors are free to publish extended versions of their work in other conferences and journals.

Submission Type

We welcome works of three different formats:

Early/Work-in-Progress Research
Extended Abstract of Completed Research
Position/Opinion Papers

Topics of Interest

Submissions should focus on AI for Systems, spanning computer architecture, programming languages, and operating systems, including but not limited to the following areas:

Computer Architecture

Agent-driven microarchitecture design, tuning, and exploration
Agent-driven chip design, hardware code generation, synthesis, and place-and-route
Agent-driven design space exploration and architectural trade-off analysis
Agent-optimized processor, accelerator, and heterogeneous system design
GPU, TPU, and accelerator architectures and kernel optimization via autonomous agents
Memory systems, cache hierarchies, interconnects, and storage optimization using agentic AI
Agent-based performance, power, energy, and reliability modeling
Agentic workflows and surrogate models to accelerate architectural simulation and evaluation
Autonomous methods for identifying architectural bottlenecks and inefficiencies

Systems

Agent-assisted compilers, program analysis, and code generation
Automatic kernel transformation, scheduling, and tuning
Cross-layer co-design across compilers, runtimes, operating systems, and hardware via agentic workflows
Agent-driven scheduling, resource management, and system-level optimization
Intelligent and autonomous runtime systems for heterogeneous and accelerator-rich platforms
Agent-driven memory management, caching, and I/O policies

Infrastructure and Evaluation

Datasets tailored for agentic hardware and systems evaluation
Benchmarks for assessing agent-driven system workflows
Open-source infrastructure enabling agent-driven approaches across the architecture–language–OS stack

Reviewing Details

Submission and Platform: All submissions will be handled through the OpenReview platform. The review process will be single-blind; therefore, submissions should include author names and affiliations and should not be anonymized.

Decisions: All accept/reject decisions will be made exclusively by the organizing committee. Submissions will be evaluated for relevance to the event's theme, technical novelty, and clarity of presentation.

Submission Details

Submission deadline: May 19, 2026 (AoE) (Deadline Extended)

Author Notification: May 22, 2026

Paper length: (This is not a strict limit, authors are encouraged to adhere to it if possible.)

Early/Work-in-Progress Research: 4 pages
Extended Abstract of Completed Research: 2 pages (with pointer to full-length paper)
Position/Opinion Papers: 2 pages

Format: Standard ISCA 2026 Latex Template

OpenReview Submission Site