Why should I run this skill before starting a scientific task?

Running this skill prevents resource exhaustion and system crashes by recommending the best computational strategy based on your actual hardware, such as using chunked data loading for large files.

Does it support GPU detection on Mac?

Yes, it specifically detects Apple Silicon (M1/M2/M3/M4) chips and recommends Metal-accelerated backends like PyTorch MPS or TensorFlow-Metal.

Can I use the output of this skill directly in my code?

Yes, the skill generates a .claude_resources.json file that can be easily loaded into Python scripts to dynamically set parameters like n_jobs or device types.

Are there any external dependencies required?

The underlying detection script requires the 'psutil' Python package, which can be installed via 'uv pip install psutil' or standard pip.

What happens if no GPU is detected?

The skill will identify the lack of GPU acceleration and provide alternative recommendations for CPU-optimized libraries and sequential or parallel processing strategies.

System Resource Profiler

Name: System Resource Profiler
Author: Microck

byMicrock

•

数据科学与机器学习

Analyzes hardware capabilities and provides strategic recommendations for computationally intensive scientific tasks.

The System Resource Profiler is designed to optimize scientific computing workflows by identifying available hardware resources before executing heavy tasks. It automatically detects CPU cores, GPU availability (NVIDIA, AMD, and Apple Silicon), RAM constraints, and disk space to generate a strategic JSON report. This report informs critical architectural decisions, such as whether to use parallel processing with Dask, GPU acceleration with PyTorch, or memory-efficient out-of-core computing, ensuring your code runs efficiently within the limits of your environment.

主要功能

01Real-time monitoring of available RAM and disk space with usage thresholds

0281 GitHub stars

03Comprehensive detection of CPU cores, architecture, and logical processors

04Multi-vendor GPU identification for NVIDIA (CUDA), AMD (ROCm), and Apple Silicon (Metal)

05Context-aware recommendations for parallel processing libraries like joblib and Dask

06Hardware-specific acceleration guidance for machine learning frameworks like PyTorch and JAX

使用场景

01Optimizing worker counts and parallelization strategies for large-scale data processing

02Selecting the correct hardware backend (CUDA, ROCm, or MPS) before training machine learning models

03Determining if datasets require out-of-core processing techniques based on available memory

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add microck/ordinary-claude-skills get-available-resources

For use in Claude.ai and ChatGPT

Download Skill