Skip to main content 首页 创作者 hiyenwong ai_collection chia-agentic-hardware-software-codesign
chia-agentic-hardware-software-codesign CHIA framework for principled agentic AI-driven hardware/software co-design. Treat hardware/software design flows as directed cyclic graphs (CHIA loops) with nodes executing SoC design tools, simulators, AI models, and evolutionary agents. Supports Chipyard, gem5, ChampSim, FireSim, Vivado, AlphaEvolve, and more. Enables isolation, profiling, fault-tolerant execution, and reliability across heterogeneous systems.
跳到安装 Skills Marketplace 发现并探索由社区构建的 Agent Skills
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/hiyenwong/ai_collection --skill chia-agentic-hardware-software-codesign命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
下载 Zip 下载中... name chia-agentic-hardware-software-codesign description CHIA framework for principled agentic AI-driven hardware/software co-design. Treat hardware/software design flows as directed cyclic graphs (CHIA loops) with nodes executing SoC design tools, simulators, AI models, and evolutionary agents. Supports Chipyard, gem5, ChampSim, FireSim, Vivado, AlphaEvolve, and more. Enables isolation, profiling, fault-tolerant execution, and reliability across heterogeneous systems. activation_keywords ["CHIA","agentic AI","hardware software co-design","computer architecture","systems engineering","RTL generation","gem5 simulation","Chipyard","FireSim","Vivado","AlphaEvolve","evolutionary agents","microarchitectural optimization","IPC optimization","critical path optimization","GitHub issue automation","硬件软件协同设计","智能体驱动设计","系统工程","计算机架构"] arxiv_id 2606.2735 authors Angela Cui, Ferran Hermida-Rivera, Jack Toubes, Raghav Gupta, Jim Fang, Chengyi Lux Zhang, Ella Schwarz, Junha Kim, Yakun Sophia Shao, Borivoje Nikolic, Christopher W. Fletcher, Sagar Karandikar published 2026-06-25T00:00:00.000Z categories ["cs.AR","cs.AI","cs.SE"]
CHIA: Agentic AI-Driven Hardware/Software Co-Design Framework
Core Contribution
CHIA (Agentic AI Hardware/Software Co-design) is an open-source framework that enables principled research on AI-driven hardware/software co-design. It treats the construction and deployment of co-design flows as a first-class objective.
Key Innovation: CHIA Loops
CHIA expresses agentic AI-driven hardware and software design flows as directed cyclic graphs called CHIA loops . Each node executes:
System-on-chip design tools (Chipyard, Hammer, Vivado)
Microarchitectural simulators (gem5, ChampSim)
Software build systems
AI models (LLMs, evolutionary agents)
Evolutionary coding agents (AlphaEvolve, AdaEvolve)
Framework Components
1. CHIA Library
Provides node implementations for popular tools:
Chipyard : Agile SoC design frameworkgem5 : Architectural simulationChampSim : Cache simulatorFireSim : FPGA-accelerated simulationHammer : Physical design (commercial CAD tools)Vivado : FPGA synthesisAlphaEvolve : LLM-based evolutionary agentsAdaEvolve : Adaptive evolutionary agents
2. Principled Science Features
Isolation : Separates AI models from hardware toolsProfiling : Performance monitoring mechanismsFault tolerance : Robust execution handlingScalability : Reliability across hundreds of heterogeneous systemsMulti-platform : CPUs, FPGAs, GPUs across cloud/on-prem
Case Studies
1. RTL-to-gem5 Simulator Alignment
Automatic generation and alignment of RTL with gem5 simulators for consistency verification.
2. LLM-Driven Microarchitectural Feature Implementation
Agentic AI automatically implements microarchitectural features in RTL code.
3. IPC-Aware Critical Path Optimization
Intelligent optimization of instruction-level parallelism aware of critical paths.
4. Evolutionary Architectural Discovery
Evolutionary agents discover novel architectural configurations.
5. Maintainer-Friendly GitHub Issue Fixing
Agentic automation for resolving hardware/software design issues.
Implementation Patterns
Creating a CHIA Loop
from chia import Loop, Node, Edge
loop = Loop(
nodes=[
Node("chipyard" , tool="chipyard" , config="soconfig.yaml" ),
Node("gem5_sim" , tool="gem5" , workload="benchmark_suite" ),
Node("llm_agent" , model="claude-3" , task="rtl_optimization" ),
Node("alphaevolve" , strategy="evolutionary" ),
],
edges=[
Edge("chipyard" , "gem5_sim" , "rtl_output" ),
Edge("gem5_sim" , "llm_agent" , "perf_metrics" ),
Edge("llm_agent" , "alphaevolve" , "optimization_hints" ),
Edge("alphaevolve" , "chipyard" , "design_candidates" ),
]
)
result = loop.run(
isolation=True ,
profiling=True ,
fault_tolerance="retry" ,
max_retries=3
)
Node Configuration
node:
name: gem5_simulation
tool: gem5
config:
cpu_model: O3CPU
cache_hierarchy: MESI_two_level
memory: DDR4_2400
workload:
type: spec2017
benchmarks: [600. perlbench , 603. bwaves ]
output:
format: json
metrics: [ipc , cache_miss_rate , branch_mispredict_rate ]
System Engineering Principles
1. Directed Cyclic Graphs
Composition : Nodes can be chained in cyclesIteration : Feedback loops for iterative optimizationParallelization : Independent nodes execute concurrently
2. Tool Abstraction
Input/output formats : Standardized interfacesExecution semantics : Fault-tolerant executionResource management : CPU/GPU/FPGA allocation
3. Agent Integration Supports multiple agent types:
LLM-based : Code generation, design explorationEvolutionary : Architectural search, parameter optimizationReinforcement learning : Policy optimization
Research Methodology
Experiment Design
from chia.experiment import Experiment
exp = Experiment(
loop=my_chia_loop,
isolation_level="tool_agent" ,
profiling=["latency" , "memory" , "gpu_util" ],
seed=42 ,
repetitions=5
)
results = exp.run_multi_seed(seeds=[42 , 123 , 456 , 789 , 1024 ])
Fault Tolerance Patterns
from chia.fault_tolerance import RetryHandler
handler = RetryHandler(
strategy="exponential_backoff" ,
max_retries=5 ,
initial_delay=2.0 ,
max_delay=60.0
)
node.configure(handler=handler)
Deployment Architecture
Heterogeneous System Support
Cloud : AWS, GCP, AzureOn-prem : Local clustersFPGA : FireSim infrastructureGPU : CUDA nodes
Reliability Patterns
Checkpointing : State persistence across failuresHot-swap : Replace nodes without stopping loopMonitoring : Real-time health checks
Applications
Computer Architecture Research
Automated design space exploration
Microarchitectural optimization
Performance modeling
Compiler Optimization
LLVM pass optimization
Code generation strategies
VLSI Design
RTL generation
Physical design flows
DRC checking
Open Source Stack
Code : github.com/chia-framework/chiaDocumentation : chia-framework.orgNodes : Pre-built implementationsExamples : Case study implementations
Pitfalls
1. Tool Version Compatibility Different versions may produce incompatible outputs. Always specify exact tool versions in node config.
2. Agent Prompt Design LLM agents require precise prompts for hardware design tasks. Use domain-specific templates.
3. Resource Allocation Gem5 simulations consume significant memory. Profile node resource requirements before deployment.
4. Loop Deadlocks Cyclic graphs can deadlock if nodes wait indefinitely. Implement timeout mechanisms.
5. Simulation-Real Gap FireSim FPGA results may differ from gem5. Validate across multiple platforms.
Related Skills
agent-first-bootstrap: Agent-first methodologyhardware-aware-quantum-compilation: Hardware-aware compilationagentic-scientific-workflow: Agentic research workflowsdistributed-quantum-control-architecture: Distributed control architecture
References
