accelforge
AccelForge
Decision gist · record as of 2026-08-14
Yes, if you are designing or researching tensor accelerators and need a framework for full-stack modeling and workload optimization. The package is actively maintained, has low installation friction, and integrates well with the HWComponents ecosystem. However, verify the license terms first, and be prepared for a substantial dependency footprint. Not suitable for general-purpose tensor computation—use it for accelerator design and exploration, not as a replacement for PyTorch or TensorFlow.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python >= 3.8.
- The hwcomponents backend and its specialized variants (adc, cacti, library, neurosim) may have additional system or environment requirements not documented in this package's metadata.
- Low friction installation with a pure-Python wheel.
License · maintenance · safety
(unclear) — License status is unclear—no SPDX identifier or raw license text is available in the package metadata. Before adopting this package in a commercial or restricted context, verify the actual license terms from the project repository or maintainers.
last release 2026-07-14 (31 days)
0 known vulnerabilities (OSV.dev, 2026-08-14) · 92,023 downloads/mo, #13,487 on PyPI
Alternatives
Verify before relying
pip install accelforge
import accelforge
# See examples/ and notebooks/ directories in the repository for architecture and workload specifications- Exact license type and any redistribution or commercial-use restrictions
- Whether hwcomponents and its variants require external system dependencies or configuration
- Performance characteristics and scalability limits for large accelerator designs
- Maturity level and API stability guarantees for the modeling framework
What it is and what it does
AccelForge is a Python framework for modeling and optimizing tensor algebra accelerators. It provides a full-stack modeling environment where you define hardware architectures (with multiple compute unit types), tensor workloads (expressed as Einsums), and mappings between them, then uses the HWComponents backend to estimate area, energy, power, and throughput costs. The framework handles fusion-aware optimization, meaning it can optimize across cascades of tensor operations to find the best end-to-end performance and energy efficiency.
The package is built on established scientific Python libraries (numpy, pandas, sympy, symengine) for numerical and symbolic computation, Pydantic for input validation, and visualization tools (matplotlib, plotly) for design exploration. It's intended for researchers and hardware engineers designing accelerators, optimizing tensor workloads, or exploring the design space of specialized compute hardware.
Use it for
- Design and evaluate custom tensor accelerator architectures for deep learning or scientific computing workloads
- Optimize tensor operation mappings and fusion strategies to minimize energy and area on fixed hardware
- Explore trade-offs between compute unit types, memory hierarchies, and interconnect topologies
- Validate accelerator designs against real workloads before hardware implementation
- Automate accelerator design space exploration and sensitivity analysis via Python scripting
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you are designing or researching tensor accelerators and need a framework for full-stack modeling and workload optimization.
The package is actively maintained, has low installation friction, and integrates well with the HWComponents ecosystem. However, verify the license terms first, and be prepared for a substantial dependency footprint. Not suitable for general-purpose tensor computation—use it for accelerator design and exploration, not as a replacement for PyTorch or TensorFlow.
Install
accelforge on PyPI
Before you install
Low friction installation with a pure-Python wheel. Active maintenance as of late July 2026, though repository metadata is not publicly available. The 22 runtime dependencies include established libraries (numpy, pandas, pydantic) and specialized hardware modeling packages (hwcomponents suite), which may require careful environment setup.
Requires Python >= 3.8. The hwcomponents backend and its specialized variants (adc, cacti, library, neurosim) may have additional system or environment requirements not documented in this package's metadata.
License in practice
License status is unclear—no SPDX identifier or raw license text is available in the package metadata. Before adopting this package in a commercial or restricted context, verify the actual license terms from the project repository or maintainers.
Quickstart
pip install accelforge
import accelforge
# See examples/ and notebooks/ directories in the repository for architecture and workload specificationsVerify before relying
- Exact license type and any redistribution or commercial-use restrictions
- Whether hwcomponents and its variants require external system dependencies or configuration
- Performance characteristics and scalability limits for large accelerator designs
- Maturity level and API stability guarantees for the modeling framework
Package facts
| License | Not declared unclear |
| Python support | Supports the current Python release >=3.8 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 22 packagesnumpypandastqdmpydanticpydantic_coreruamel.yamljinja2islpy-barvinoksympysymengineparetosetmatplotlibplotlypydotplatformdirsjoblibrequestshwcomponentshwcomponents-adchwcomponents-cactihwcomponents-libraryhwcomponents-neurosim |
| Maintenance | Actively maintained 31 days since the last release |
| First released | |
| Downloads | 92,023 / month, #13,487 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: accelforge-1.0.467-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “tensor accelerator modeling”
- accelforgeAccelForge models, designs, and explores tensor algebra accelerators…
- tokamaxTokamax provides custom accelerator kernels for JAX, including…
- accelerateAccelerate abstracts away distributed training boilerplate for…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also cloud-accelerator-diagnostics · transformer-engine-cu13 · transformer-engine-cu12 · transformer-engine · tensorflow-intel · opt-einsum · fa3-fwd · nvidia-cusparse · nvidia-cusparse-cu12 · spandrel-extra-arches