| name | hipgraph-analytics |
| description | Leverages GPU acceleration to process and analyze complex graph structures using hipGRAPH. Note: hipGRAPH is early access. |
Purpose
Utilize hipGRAPH for high-performance graph processing and analysis, while acknowledging its early-access status.
When to use
- When a data science task involves complex network or graph structures.
- To execute diverse graph algorithms such as centrality, traversal, similarity, sampling, and labeling.
- When graph outputs need to integrate seamlessly with hipDF DataFrames across the ROCm-DS ecosystem.
When not to use
- Production Environments: The hipGRAPH libraries are currently in an early access state. Running production workloads with these libraries is explicitly not recommended by AMD.
- If the workload is purely tabular (use
hipdf-pandas-port) or purely vector search (use hipvs-ann).
Procedure
- Verify that the requested workload is a non-production, experimental, or benchmark task.
- Formulate the graph using hipGRAPH APIs, loading data (potentially from hipDF).
- Execute the target algorithm (e.g., centrality, traversal, PageRank, BFS, node2vec).
- Analyze the results or pipe them back into hipDF.
- Report on the execution, highlighting any early-access instability or API quirks.
