Compilers & kernel DSLs (23)
Compilers, DSLs and code-generation stacks for inference.
MAX and Mojo (Modular)
compiler-frameworkhighmixedcompilers/w7k-modular-max-mojo
draftmedium · 0.70
Qualcomm AI Engine Direct (QNN) SDK
compiler-frameworkvery-highaotcompilers/w7k-qualcomm-qnn
draftlow · 0.50
AMD Composable Kernel (CK)
template-libraryhightemplate-instantiationcompilers/w7k-amd-composable-kernel
verifiedhigh · 0.85
Apache TVM
compiler-frameworkhighaotcompilers/apache-tvm
verifiedhigh · 0.85
Arm Kleidi Libraries (KleidiAI)
template-librarylowtemplate-instantiationcompilers/w7k-arm-kleidiai
verifiedmedium · 0.75
CANN / Ascend GraphEngine (GE)
graph-optimizervery-highgraph-rewritecompilers/w7k-cann-ascend-graphengine
verifiedmedium · 0.75
Cerebras CSL compiler (cslc) and the CSL kernel language
kernel-dslhighaotcompilers/w7k-cerebras-cslc
verifiedmedium · 0.80
CUDA NVRTC (runtime compilation library)
compiler-frameworkhighjitcompilers/w7k-cuda-nvrtc
verifiedhigh · 0.85
CUTLASS / CuTe
template-libraryvery-hightemplate-instantiationcompilers/cutlass-cute
verifiedhigh · 0.90
IREE
compiler-frameworkhighaotcompilers/iree
verifiedmedium · 0.80
llama.cpp SYCL backend (hand-written kernels over oneDNN, oneMKL and joint_matrix)
kernel-dslhighjitcompilers/llama-cpp-sycl-backend
verifiedmedium · 0.80
MLIR
compiler-frameworkvery-highmixedcompilers/mlir
verifiedhigh · 0.90
nvFuser (MosaicML fusion compiler, now under NVIDIA)
graph-optimizerhighjitcompilers/w7k-nvfuser
verifiedmedium · 0.80
ONNX Runtime
runtimelowgraph-rewritecompilers/onnx-runtime
verifiedhigh · 0.85
OpenVINO
runtimelowgraph-rewritecompilers/openvino
verifiedmedium · 0.80
Pallas
kernel-dslmoderatejitcompilers/pallas
verifiedhigh · 0.90
TensorRT
compiler-frameworkhighaotcompilers/w7k-tensorrt
verifiedhigh · 0.85
ThunderKittens (and the HipKittens AMD port)
template-libraryhightemplate-instantiationcompilers/w7k-thunderkittens
verifiedmedium · 0.80
TileLang (Tile Language)
kernel-dslmoderatemixedcompilers/w7k-tilelang
verifiedmedium · 0.80
torch.compile / TorchInductor
compiler-frameworkmoderatejitcompilers/torch-compile-torchinductor
verifiedhigh · 0.85
Triton
dslmoderatejitcompilers/triton
verifiedhigh · 0.90
TT-Forge (Tenstorrent's MLIR-based compiler; the TT-Bridge / graph-frontend layer)
compiler-frameworkhighaotcompilers/w7k-tt-forge
verifiedmedium · 0.75
XLA
compiler-frameworkhighaotcompilers/xla
verifiedhigh · 0.85
Most referenced
The records other records point at. Useful as an entry point: a heavily referenced accelerator or engine is one many decisions in this repo assume.
| Record | Referenced by | Cites |
|---|---|---|
| CUTLASS / CuTe | 1 | 2 |
| torch.compile / TorchInductor | 1 | 2 |
| Triton | 1 | 2 |