alibaba/BladeDISC
BladeDISC is an end-to-end DynamIc Shape Compiler project for machine learning workloads.
When running machine learning models, especially those with varying input sizes (dynamic shapes), their performance can be slow on standard setups. This project takes your existing TensorFlow or PyTorch models and optimizes them to run much faster on GPUs and CPUs. It's designed for machine learning engineers and MLOps professionals who want to deploy high-performing models in production.
919 stars. No commits in the last 6 months.
Use this if you need to significantly speed up the inference or training of your TensorFlow or PyTorch machine learning models, particularly those with dynamic input shapes, on various hardware like NVIDIA, AMD, or Hygon GPUs, and x86/AArch64 CPUs.
Not ideal if your machine learning workloads always use static input shapes and are already well-optimized with existing static compilers, or if you are not working with TensorFlow or PyTorch.
Stars
919
Forks
169
Language
C++
License
Apache-2.0
Category
Last pushed
Dec 30, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/alibaba/BladeDISC"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
apache/tvm
Open Machine Learning Compiler Framework
uxlfoundation/oneDNN
oneAPI Deep Neural Network Library (oneDNN)
Tencent/ncnn
ncnn is a high-performance neural network inference framework optimized for the mobile platform
OpenMined/TenSEAL
A library for doing homomorphic encryption operations on tensors
iree-org/iree-turbine
IREE's PyTorch Frontend, based on Torch Dynamo.