SparseTIR: Sparse Tensor Compiler for Deep Learning
-
Updated
Mar 31, 2023 - Python
#
tensor-compiler
Here are 14 public repositories matching this topic...
Experimental tensor compiler with a Python frontend, typed SSA IR, semantics preservation tests, and Metal lowering.
-
Updated
Sep 24, 2026 - Python
Optimizing RetinaFace inference using TVM.
-
Updated
Feb 27, 2023 - C++
A tiny educational tensor compiler built from scratch to explain the core ideas behind Apache TVM.
-
Updated
Oct 12, 2026 - Python
Tiny tensor compiler: one tensor program compiled to a GPU simulator (tiny-gpu) and a systolic-array TPU (tiny-tpu), checked bit-exact against numpy.
react compiler webassembly rtl astro matrix-multiplication int8 verilator tpu pyodide simt deep-learning-compiler tensor-compiler systolic-array gpu-simulator compiler-education
-
Updated
Oct 27, 2026 - Python
Tensor-aware meta-tracing JIT
-
Updated
Oct 21, 2026 - Python
SchedForge — A Target-Aware CPU Tensor Compiler. Forge tensor schedules into efficient CPU code.
-
Updated
Aug 20, 2026 - C++
Verified C++20 tensor compiler/runtime with bounded graph execution, Mul/ReGLU workloads, independent verification, and reproducible evidence.
compiler reproducible-research inference systems-programming softmax graph-rewriting cpp20 machine-learning-systems numerical-computing tensor-compiler execution-runtime memory-planning reglu
-
Updated
Aug 10, 2026 - C++
Tensor compiler: PyTorch model -> graph fusion -> loop-nest IR -> tiling/AVX2/OpenMP C and CUDA shared-memory kernels, with autotuning and roofline analysis. In active development.
-
Updated
Oct 11, 2026
A tensor compiler that writes its own ARM64 machine code. Pure Python standard library — no LLVM, no assembler, no numpy. Every instruction verified against Apple's own assembler.
python compiler neon jit simd code-generation arm64 no-dependencies apple-silicon tensor-compiler greenai-solutions
-
Updated
Aug 16, 2026 - Python
Tensor graph compiler in Rust: bit-exact CPU backends, ONNX import and an optional wgpu GPU backend (Vulkan, Metal, DX12). Every claim is tested and measured; GPU numerics are documented.
rust neural-network compiler gpu property-based-testing model-checking tensor intermediate-representation compiler-optimization shape-inference dead-code-elimination common-subexpression-elimination kani wgpu numerical-computing wgsl constant-folding graph-compiler tensor-compiler ml-compiler
-
Updated
Oct 31, 2026 - Rust
A small ahead-of-time compiler for tensor expressions, written in C++20. It parses a simple language, infers shapes, builds an SSA graph IR, and runs passes like constant folding, operator fusion, and buffer reuse. It then generates C++ and CUDA code, with benchmarks comparing fused and unfused output.
-
Updated
Nov 3, 2026 - C++
Tensor compiler using JIT kernel generation for ARM processors
-
Updated
Oct 4, 2026 - C++
A tensor compiler for the transformer nanoinfer runs by hand: graph IR with symbolic shapes, fusion, schedules and C codegen, checked against nanoinfer
python compiler transformers inference code-generation from-scratch kernel-fusion llm tensor-compiler
-
Updated
Oct 26, 2026 - Python
Add this topic to your repo
To associate your repository with the tensor-compiler topic, visit your repo's landing page and select "manage topics."