This is an automated email from the ASF dual-hosted git repository. spectrometerHBH pushed a commit to branch docs/refresh-readme in repository https://gitbox.apache.org/repos/asf/tvm-tirx.git
commit 2bb73c9d79c5c8ed681fbd1078f6ea770ba60ce9 Author: spectrometerHBH <[email protected]> AuthorDate: Tue Nov 3 17:19:18 2026 -0500 [DOCS] Refresh README positioning and features --- README.md | 25 ++++++++++++------------- 1 file changed, 12 insertions(+), 13 deletions(-) diff --git a/README.md b/README.md index 3bd023ff4..4042aea4c 100644 --- a/README.md +++ b/README.md @@ -15,25 +15,24 @@ <!--- specific language governing permissions and limitations --> <!--- under the License. --> -# TVM TIRx: ML Compiler Foundation for the Agentic Era +# TVM TIRx 📚 [Documentation](https://tvm.apache.org/tirx/) | 🚀 [Quick Start](https://tvm.apache.org/tirx/get_started/quick_start.html) -Apache TVM TIRx is a minimal, open ML compiler foundation for the agentic era. +**We believe that, in the future, all GPU kernels will be developed, optimized, and verified by AI agents.** + +And we believe they'll need a **minimal, stable, and extensible DSL and compiler foundation** to power agent-driven kernel development loops. That's what we're building with TIRx. + +But wait, **why would agents need DSLs and compilers at all?** We've been asking ourselves the same question. Read our [blog post](#) for our take. ## Features -- **Minimal, stable core IR** - a small, stable set of core AST nodes, designed for common needs and for extension. -- **Predictable utilities** - TVM-FFI bridge generation, symbolic integer analysis, structural traversal and mapping, and minimal code generation for CUDA and host LLVM. -- **Dynamic extension by design** - all compiler APIs are open, so developers and agents can iterate quickly on extensions in separate packages that link to TIRx dynamically, without rebuilding one monolithic bundle. -- **First-class cross-language support** - Python, C++, and native Rust. Prototype an analysis or extension in Python, then bake it into Rust or C++; extensions work across all supported languages. -- **Built on stable toolchains** - a minimal layer for ML compilation needs that connects to stable compiler toolchains and leverages their backend capabilities when possible. +- **Minimal, stable core IR** — a small, stable set of core AST nodes, designed for common needs and extensions. +- **Predictable utilities** — TVM-FFI bridge generation, symbolic integer analysis, structural traversal and mapping, and minimal code generation for CUDA and host LLVM. +- **Dynamic extension by design** — agents can build and iterate on extensions in independent packages, without rebuilding a monolithic compiler. +- **First-class cross-language support** — Python, C++, and Rust, allowing agents to prototype rapidly and move performance-critical components across languages. +- **Built on stable toolchains** — leverage established compiler backends instead of reinventing them. ## Documentation -Check out the [documentation site](https://tvm.apache.org/tirx/) for the full documentation, including: - -- [Quick Start](https://tvm.apache.org/tirx/get_started/quick_start.html) -- [IR Guides](https://tvm.apache.org/tirx/ir_guides/overview.html) -- [Programming Guides](https://tvm.apache.org/tirx/programming_guides/index.html) -- [Extending TIRx](https://tvm.apache.org/tirx/advanced/extending_ir_in_rust.html) +Check out the [documentation](https://tvm.apache.org/tirx/) to get started.
