# CUTLASS

## Docs

- [cutlass::Array](https://mintlify.wiki/NVIDIA/cutlass/api/cpp/array.md): Statically sized array for all CUTLASS-supported numeric types, safe for use in unions
- [Epilogue Operators](https://mintlify.wiki/NVIDIA/cutlass/api/cpp/epilogue-operators.md): Output transformation operators for GEMM epilogues
- [cutlass::gemm::device::Gemm](https://mintlify.wiki/NVIDIA/cutlass/api/cpp/gemm.md): Device-level GEMM operator template for general matrix-matrix multiplication
- [cutlass::layout Types](https://mintlify.wiki/NVIDIA/cutlass/api/cpp/layouts.md): Layout functions mapping logical coordinates to linear memory
- [cutlass::TensorRef](https://mintlify.wiki/NVIDIA/cutlass/api/cpp/tensor-ref.md): Template for objects pointing to tensors of arbitrary rank and layout within memory
- [cute::Layout](https://mintlify.wiki/NVIDIA/cutlass/api/cute/layout.md): CuTe Layout type for mapping logical coordinates to linear indices
- [cute::Tensor](https://mintlify.wiki/NVIDIA/cutlass/api/cute/tensor.md): CuTe Tensor type combining data storage with multi-dimensional layout
- [cute::TiledCopy](https://mintlify.wiki/NVIDIA/cutlass/api/cute/tiled-copy.md): CuTe TiledCopy for partitioning copy operations across threads
- [cute::TiledMma](https://mintlify.wiki/NVIDIA/cutlass/api/cute/tiled-mma.md): CuTe TiledMma for partitioning matrix multiply operations across threads
- [Atom Operations](https://mintlify.wiki/NVIDIA/cutlass/api/dsl/atoms.md): CuTe DSL atom operations for copy and matrix multiply-accumulate
- [Kernel Definition](https://mintlify.wiki/NVIDIA/cutlass/api/dsl/kernel-definition.md): CuTe DSL kernel and JIT function decorators for GPU programming
- [Memory Operations](https://mintlify.wiki/NVIDIA/cutlass/api/dsl/memory-operations.md): CuTe DSL memory management and tensor creation APIs
- [DSL Operations](https://mintlify.wiki/NVIDIA/cutlass/api/dsl/operations.md): CuTe DSL operations for GEMM, copy, and tensor manipulation
- [cutlass.op.Gemm](https://mintlify.wiki/NVIDIA/cutlass/api/python/gemm.md): Python API reference for General Matrix Multiply (GEMM) operations
- [Layout Types Reference](https://mintlify.wiki/NVIDIA/cutlass/api/python/layout-types.md): Memory layout types used in CUTLASS Python operations
- [Python Operations Overview](https://mintlify.wiki/NVIDIA/cutlass/api/python/operations.md): Overview of CUTLASS Python operations and the high-level API
- [Utility Functions](https://mintlify.wiki/NVIDIA/cutlass/api/python/utilities.md): Helper functions and utilities for CUTLASS Python operations
- [Ampere Architecture (SM80/86)](https://mintlify.wiki/NVIDIA/cutlass/architectures/ampere.md): CUTLASS support for NVIDIA Ampere architecture GPUs
- [Blackwell Architecture (SM100/120)](https://mintlify.wiki/NVIDIA/cutlass/architectures/blackwell.md): CUTLASS support for NVIDIA Blackwell architecture with UMMA and TMEM
- [Hopper Architecture (SM90)](https://mintlify.wiki/NVIDIA/cutlass/architectures/hopper.md): CUTLASS support for NVIDIA Hopper architecture GPUs with WGMMA and TMA
- [Architecture Support Overview](https://mintlify.wiki/NVIDIA/cutlass/architectures/overview.md): Overview of NVIDIA GPU architectures supported by CUTLASS
- [CuTe Library](https://mintlify.wiki/NVIDIA/cutlass/concepts/cute.md): Understanding CuTe - the core tensor abstraction library powering CUTLASS 3.x
- [GEMM Operations](https://mintlify.wiki/NVIDIA/cutlass/concepts/gemm.md): Understanding General Matrix Multiplication and hierarchical decomposition in CUTLASS
- [Memory Layouts](https://mintlify.wiki/NVIDIA/cutlass/concepts/layouts.md): Understanding memory layout strategies for optimal GPU performance in CUTLASS
- [Tensor Core Programming](https://mintlify.wiki/NVIDIA/cutlass/concepts/tensor-cores.md): Understanding and programming NVIDIA Tensor Cores for accelerated matrix operations
- [Convolution Operations](https://mintlify.wiki/NVIDIA/cutlass/cpp/convolution.md): CUTLASS convolution API for efficient 2D and 3D convolutions
- [CuTe Library](https://mintlify.wiki/NVIDIA/cutlass/cpp/cute-library.md): CuTe - A modern C++ template library for tensor operations in CUDA
- [Epilogue Operations](https://mintlify.wiki/NVIDIA/cutlass/cpp/epilogue.md): Customize output operations with CUTLASS epilogue functors
- [GEMM API](https://mintlify.wiki/NVIDIA/cutlass/cpp/gemm-api.md): General Matrix Multiply (GEMM) API reference and usage guide
- [C++ API Overview](https://mintlify.wiki/NVIDIA/cutlass/cpp/overview.md): Overview of the CUTLASS C++ template library and core APIs
- [Basic GEMM](https://mintlify.wiki/NVIDIA/cutlass/examples/cpp/basic-gemm.md): Learn how to implement a basic General Matrix Multiply (GEMM) operation using CUTLASS
- [Batched GEMM](https://mintlify.wiki/NVIDIA/cutlass/examples/cpp/batched-gemm.md): Perform multiple independent GEMM operations efficiently using CUTLASS batched kernels
- [Convolution](https://mintlify.wiki/NVIDIA/cutlass/examples/cpp/convolution.md): Implement high-performance 2D convolution using CUTLASS Tensor Cores
- [Fused Operations](https://mintlify.wiki/NVIDIA/cutlass/examples/cpp/fused-operations.md): Combine GEMM with element-wise operations like bias addition and ReLU activation
- [Attention Kernel Example](https://mintlify.wiki/NVIDIA/cutlass/examples/dsl/attention.md): Implement Flash Attention v2 using CuTe DSL for efficient transformer inference
- [GEMM Kernel in CuTe DSL](https://mintlify.wiki/NVIDIA/cutlass/examples/dsl/gemm-kernel.md): Build a complete dense GEMM kernel using CuTe DSL with tensor core operations
- [Mixed Precision GEMM](https://mintlify.wiki/NVIDIA/cutlass/examples/dsl/mixed-precision.md): Implement mixed-precision matrix multiplication with automatic type conversion and scaling
- [Custom Epilogue Example](https://mintlify.wiki/NVIDIA/cutlass/examples/python/custom-epilogue.md): Fuse custom operations into GEMM epilogues for improved performance
- [Python GEMM Example](https://mintlify.wiki/NVIDIA/cutlass/examples/python/gemm-python.md): Learn how to implement matrix multiplication using CUTLASS Python interface
- [Grouped GEMM Example](https://mintlify.wiki/NVIDIA/cutlass/examples/python/grouped-gemm.md): Perform multiple GEMM operations with different problem sizes in a single kernel launch
- [Installation](https://mintlify.wiki/NVIDIA/cutlass/installation.md): Install CUTLASS for C++ and Python development
- [Introduction to CUTLASS](https://mintlify.wiki/NVIDIA/cutlass/introduction.md): High-performance CUDA C++ and Python template library for linear algebra operations at all levels and scales
- [Performance Benchmarks](https://mintlify.wiki/NVIDIA/cutlass/performance/benchmarks.md): CUTLASS performance benchmarks across NVIDIA GPU architectures
- [Optimization Techniques](https://mintlify.wiki/NVIDIA/cutlass/performance/optimization.md): Strategies and best practices for optimizing CUTLASS kernel performance
- [Profiling with CUTLASS Profiler](https://mintlify.wiki/NVIDIA/cutlass/performance/profiling.md): Use the CUTLASS profiler to measure and analyze kernel performance
- [CuTe DSL](https://mintlify.wiki/NVIDIA/cutlass/python/cute-dsl.md): Python DSL for writing high-performance CUDA kernels using CuTe abstractions
- [Python Examples](https://mintlify.wiki/NVIDIA/cutlass/python/examples.md): Explore CUTLASS Python examples including CuTe DSL kernels and Jupyter notebooks
- [Python DSL Overview](https://mintlify.wiki/NVIDIA/cutlass/python/overview.md): Overview of CUTLASS Python packages including the Python interface and CuTe DSL
- [PyTorch Integration](https://mintlify.wiki/NVIDIA/cutlass/python/pytorch-integration.md): Integrate CUTLASS Python kernels with PyTorch workflows
- [Python Quickstart](https://mintlify.wiki/NVIDIA/cutlass/python/quickstart.md): Get started with CUTLASS Python interface and CuTe DSL
- [Quick start](https://mintlify.wiki/NVIDIA/cutlass/quickstart.md): Get up and running with CUTLASS in minutes