# AscendNPU IR users

This project enables multiple operator programming frameworks to target the Ascend backend and provides Ascend-oriented compilation and optimization. Below are examples of languages and frameworks that have integrated or use AscendNPU IR.

## Language ecosystem

| DSL | Description |
| --- | --- |
| [Triton-Ascend](https://gitcode.com/Ascend/triton-ascend) | Enables Triton developers to quickly develop Ascend operators and migrate ecosystems |
| [TileLang-Ascend (branch npuir)](https://github.com/tile-ai/tilelang-ascend/tree/npuir) | Tile-level programming for high-performance kernels, balancing productivity and low-level optimization |
| [DLCompiler](https://github.com/DeepLink-org/DLCompiler) | Deep learning compiler extending Triton, with cross-architecture DSL extension and automatic optimization |
| [FlagTree](https://github.com/flagos-ai/flagtree) | Open-source AI compiler based on Triton with unified compilation across multiple backends |