# accelerated-computing-cudf

Guidance for using cuDF and dask-cuDF to move pandas-style data work onto GPUs and check that results keep the expected semantics.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 3100 installs · ★ 3735
- Author: NVIDIA
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/NVIDIA/skills/tree/HEAD/skills/accelerated-computing-cudf/SKILL.md
- Updated: 2026-10-23
- HTML: https://affpapa.org/en/nevibecode/skills/accelerated-computing-cudf

## What it does

The skill selects a cuDF approach based on the task: `cudf.pandas` when keeping pandas code mostly intact matters, explicit cuDF for targeted migrations and control, and dask-cuDF when a dataset exceeds GPU memory. Its guidance covers CSV and Parquet input, joins, grouped calculations, and missing-value handling.

It also walks through version compatibility, comparisons against pandas, and GPU memory choices. For distributed work, the examples configure `enable_cudf_spill=True`; running cuDF requires a supported NVIDIA GPU and a compatible driver. Small workloads can spend more time on GPU transfer and launch overhead than on computation, so benchmark representative data before judging performance.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add NVIDIA/skills --skill accelerated-computing-cudf -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/accelerated-computing-cudf/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/accelerated-computing-cudf/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add NVIDIA/skills --skill accelerated-computing-cudf -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/accelerated-computing-cudf/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add NVIDIA/skills --skill accelerated-computing-cudf -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/accelerated-computing-cudf/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install NVIDIA/skills accelerated-computing-cudf
```

2. **Manually: put the files in place**
- File: `.github/skills/accelerated-computing-cudf/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add NVIDIA/skills --skill accelerated-computing-cudf -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/NVIDIA/skills --path skills/accelerated-computing-cudf
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/accelerated-computing-cudf/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add NVIDIA/skills --skill accelerated-computing-cudf -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/NVIDIA/skills/tree/HEAD/skills/accelerated-computing-cudf/SKILL.md

## When to use it
- move pandas DataFrame operations to a GPU
- read CSV or Parquet data with cuDF
- compare pandas and cuDF results for nulls and row ordering
- process data larger than GPU memory with dask-cuDF

## Example prompts
- Convert this pandas code to cuDF and compare the outputs on a small sample.
- Set up Parquet input and a groupby in cuDF while keeping intermediate data on the GPU.
- Show how to handle a dataset larger than GPU memory with dask-cuDF and `enable_cudf_spill=True`.

## Similar
- [xlsx](https://affpapa.org/en/nevibecode/skills/xlsx.md) — skill · 187223 installs
- [google-agents-cli-observability](https://affpapa.org/en/nevibecode/skills/google-agents-cli-observability.md) — skill · 511300 installs
- [caveman-evidence-review](https://affpapa.org/en/nevibecode/skills/caveman-evidence-review.md) — skill · 104500 installs
- [migration](https://affpapa.org/en/nevibecode/skills/migration.md) — skill · 103900 installs
- [performance-optimization](https://affpapa.org/en/nevibecode/skills/performance-optimization.md) — skill · 65200 installs
- [supabase](https://affpapa.org/en/nevibecode/skills/supabase.md) — skill · 8 installs

From the same set: [nvidia/skills](https://affpapa.org/en/nevibecode/by/nvidia/skills.md) — 178 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md