Continuous Benchmarks

LLview Continuous Benchmark interface, showing OSU benchmark results for JEDI

LLview makes it simple to track performance over time by:

• Automatic result collection
  Pull benchmark outputs (for example, CSV files) directly from your GitLab repository or any watched folder.

• Customizable summary tables
  Choose which parameters (problem size, thread count, compiler flags, etc.) define rows and columns. LLview then generates a complete table of every combination for you—no manual work.

• Multi‑curve performance graphs
  Assign any parameter to its own line or curve. LLview automatically plots all combinations so you can compare, for example, different MPI ranks, compiler versions, or hardware types at a glance.

• Embedded metadata
  Include extra context such as job IDs, Git commit hashes, build timestamps or any custom field right in your graph labels, when that information is available.

• Fully configurable style
  Control colors, line styles, axis labels, legends and more through your YAML settings to match your team’s branding and readability needs.