Skip to content

Welcome to Stream

Stream is a design-space-exploration (DSE) and constraint-optimization framework for heterogeneous dataflow accelerators - systems built by combining cores that each have their own dataflow and performance model. AIE and TPU-like cores are two example core types among others.

Given a neural-network workload (ONNX) and a hardware description (YAML), Stream schedules the workload layer-fused across the cores and uses MILP (Mixed-Integer Linear Programming) - the TETRA constraint optimization - to decide tensor placement and the transfer paths between cores. Stream builds on the ZigZag framework for per-core cost estimation.

✨ What it does

  • Heterogeneous multi-core modelling - accelerators are described as a system of cores with different compute/memory capabilities, connected by links and buses.
  • Layer-fused scheduling - parts of layers can be split and co-scheduled across cores for higher utilization and lower memory traffic.
  • Constraint-optimization allocation (TETRA) - a MILP TransferAndTensorAllocator decides where each tensor lives and how it is routed.
  • Pluggable solvers - OR-Tools GSCIP (default, license-free), OR-Tools HiGHS, and Gurobi (commercial license), all behind one API.
  • Memory- and communication-aware cost model - captures data reuse, memory hierarchy, and interconnect cost.
  • Modular pipeline - the mapping process is a sequence of stages you can configure or extend.
  • Agent-friendly - a documented public API, typed IR models, and an MCP server let an AI agent drive the framework (see Using Stream with an AI agent).

🚀 Get started

git clone https://github.com/KULeuven-MICAS/stream.git
cd stream
pip install -e .

Then run the CO pipeline on a bundled workload, with an auto-generated mapping:

python scripts/main_stream_co.py \
  --hardware stream/inputs/examples/hardware/tpu_like_quad_core.yaml \
  --workload stream/inputs/testing/workload/2conv_1_8_32_32_16_32_3.onnx

See Installation and Getting Started for details, and the User Guide for the workload, hardware, and mapping input formats.

📚 Publication

A. Symons, L. Mei, S. Colleman, P. Houshmand, S. Karl and M. Verhelst, "Stream: Design Space Exploration of Layer-Fused DNNs on Heterogeneous Dataflow Accelerators", IEEE Transactions on Computers, vol. 74, no. 1, pp. 237-249, Jan. 2025. 📄 Read the paper

Developed as part of the TETRA project at KU Leuven MICAS.