Skill Claude CodeCodex
A tool for creating structured issue files for the AKG project, including bug reports, feature proposals, and tasks. It can build them from a written description or from differences between code branches.
Skill Claude CodeCodex
A tool that creates pull-request descriptions from the differences between a current Git branch and a target branch. A pull request is a proposed set of code changes for review and merging.
Skill Claude CodeCodex
A read-only pre-submission code review tool for AKG projects. It checks formatting, security-related issues, rebase conflicts, dangerous functions, and compliance with SPEC.md.
Skill Claude CodeCodex
A tool for turning Cursor agent conversation logs into reusable guidance for improving GPU kernel code. It reads JSONL records, extracts general optimization lessons, and writes them as a SKILL.md file.
Skill Claude CodeCodex
An iterative optimization workflow that creates many code candidates, tests them, keeps the better ones, and produces new variations. An evolutionary algorithm is a search method inspired by selection and mutation.
Skill Claude CodeCodex
A code-generation agent that turns an algorithm design into executable code for GPU programming languages such as CUDA, Triton, and OpenCL. It also targets NVIDIA, AMD, and Intel hardware.
Skill Claude CodeCodex
A beginner’s guide to CUDA, NVIDIA’s system for running parallel work on GPUs. It explains how GPU threads are grouped and how different kinds of memory are used.
Skill Claude CodeCodex
An algorithm-design agent for planning computation routines before they are coded. It studies data flow, complexity, parallel work, memory access, and possible optimization approaches.
Skill Claude CodeCodex
An email-writing guide with templates and advice for business, job-search, customer, team, complaint, and feedback messages.
Skill Claude CodeCodex
A guide to detecting and reporting errors in GPU code, especially CUDA programs. It covers checks during compilation-related setup, kernel launch, and execution.
Skill Claude CodeCodex
Web scraping and data extraction best practices.
Skill Claude CodeCodex
A standard workflow for generating computational operations, with separate design, coding, and verification stages. It uses code generation followed by correctness and performance checks.
Skill Claude CodeCodex
A guide to writing Triton kernels for Ascend NPUs. Triton is a Python-based language for describing parallel operations that run in blocks across the device.
Skill Claude CodeCodex
A guide to Triton, a Python-like language for writing GPU programs in blocks of data. It explains its program model, data types, masking, and compiler-managed optimizations.
Skill Claude CodeCodex
A verification agent that checks generated code for correct results, numerical stability, edge cases, and speed. Profiling means measuring where code spends time and how it uses hardware resources.
Agent
A specification for developing agents in the MindSpore AKG project's core_v2/agents directory. It describes the agent base classes, registration process, key files, and conventions.
Skill Claude CodeCodex
A guide to writing CPU operations in C++ for PyTorch, including functions compiled and loaded as extensions. It covers tensor data, memory, types, bounds, and a standard five-step kernel structure.
Skill Claude CodeCodex
A guide to optimizing programs on ARM 64-bit processors, including Apple Silicon and AWS Graviton. It covers NEON, ARM's SIMD system for processing several values in one instruction.
Skill Claude CodeCodex
AVX (Advanced Vector Extensions) 是 x86-64 的 SIMD 指令集扩展：.
Skill Claude CodeCodex
A reference manual for CUDA C, the C and C++ interface used to write programs for NVIDIA GPUs. It covers function markers, memory declarations, and kernel launch syntax.
Skill Claude CodeCodex
A beginner’s guide to CUDA C, the C/C++ programming model for running work in parallel on NVIDIA GPUs. It covers kernels, threads, blocks, grids, and the GPU’s memory types.
Skill Claude CodeCodex
A collection of runnable examples showing how to write CUDA C kernels and connect them to PyTorch. The examples use PyTorch’s `load_inline` to compile and load the C++ and CUDA code at runtime.
Skill Claude CodeCodex
A guide for improving CUDA C programs, which run computations on NVIDIA GPUs. It covers GPU memory access, thread-block sizing, numerical stability, and debugging.
Skill Claude CodeCodex
A set of three CUDA C programming patterns for running calculations on NVIDIA graphics processors: element-by-element operations, combining many values into one result, and matrix multiplication. It includes code structures and examples for common mathematical functions.