Agent Skills by donggang
Showing 187 vetted skills indexed across 1 GitHub repositories.
dongg622benchmark-baseline
Evaluate model performance and accuracy on domestic AI chips across multiple datasets.
dongg622dtk-container
Deploys 海光DCU applications in Docker containers using the 海光DTK toolkit.
dongg622du-bootstrap
Diagnose and validate Hygon system hardware and software configurations.
dongg622hygon-pkgs
Provide official download links for Hygon DCU and AI framework packages.
dongg622hyqual
Automate hardware performance and thermal benchmarking for DCU devices.
dongg622dtk-compat
Evaluate application compatibility with Hygon DTK across OS, kernel, and CUDA APIs.
dongg622dcu-kubernetes
Deploy and configure Hygon DCU components in Kubernetes clusters using Helm charts and YAML manifests.
dongg622dcu-comm-tool
Set up and diagnose 海光DCU distributed communication environments using RCCL and RDMA.
dongg622cuda-adapt
Migrate CUDA applications to the海光DCU platform with automated code conversion and API mapping.
dongg622hyqual-bw1000
Measure memory and PCIe bandwidth for BW1000 device validation.
dongg622dcu-passthrough
Configure Hygon DCU GPU passthrough for KVM virtual machines.
dongg622dtk-rccl
Configure and troubleshoot RCCL-based distributed training across Hygon DCUs.
dongg622hy-smi
Diagnose and configure Hygon GPU/DCU hardware via the hy-smi command-line tool.
dongg622dudiag
Run hardware diagnostics and performance tests on sea光DCU systems using shell scripts.
dongg622hygon-vllm
Deploy large language models on Hygon DCU hardware with Docker.
dongg622dtk-env-install
Install and configure Hygon DTK development environments across Linux distributions.
dongg622dtk-env-usage
Set up the Hygon DTK HIP environment to compile, debug, and optimize HIP applications.
dongg622mcpy
Manage MXMACA GPU devices, memory, and data transfers via Python APIs.
dongg622mxquickstart
Guide installation and run diagnostic tests for沐曦 GPU environments.
dongg622mxoob
Automate GPU server power, sensor, log, and firmware tasks via BMC/IPMI.
dongg622mctriton
Compile and run Triton GPU kernels on曦云GPU systems.
dongg622mxaiforscience
Provide GPU-accelerated tools for physics simulations, molecular modeling, and PDE solving.
dongg622mcsolver
Perform LU, Cholesky, QR, SVD, and eigenvalue computations on GPUs.
dongg622mcpytorch
Provides PyTorch GPU acceleration compatible with沐曦 platform hardware and CUDA API.