[New Tutorial] GPU-Accelerated Rule Evaluation with cuDF (CPU vs GPU Benchmark)

[saimahesh8752]

Description

This PR adds a new community tutorial demonstrating how RAPIDS cuDF can accelerate large-scale rule-based filtering and boolean masking compared to a CPU-based Pandas implementation.

Use Case

Many data and AI workflows include a post-processing step where large volumes of model or agent outputs must be validated against deterministic rules (e.g., compliance checks, routing constraints, or risk thresholds). At scale, this rule-evaluation step can become a performance bottleneck on the CPU.

This notebook focuses on isolating and benchmarking that workload using cuDF on the GPU.

Benchmark Overview (demonstrated in notebook)

• Dataset: 1,000,000 synthetic records
• Comparison: Pandas (CPU) vs cuDF (GPU)
• Measurement: Rule-evaluation time only (data transfer excluded)
• Result: ~25–30× speedup for vectorized rule evaluation on NVIDIA T4

---

Content

• omniguard_cudf_rule_evaluation_benchmark.ipynb
  Demonstrates CPU vs GPU rule evaluation using identical logic and data.

---

Checklist

• Notebook runs end-to-end without errors on NVIDIA T4
• CPU and GPU implementations apply identical rules
• Benchmark scope and exclusions are clearly documented
• All data is synthetically generated

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB