forked from mooncake-track/Mooncake
run_demo.sh drives the whole story end to end -- conformance over every backend, the probe / routing / cost-model / plugin diagnostics, the CXL/NUMA tier benchmark, and the GPU NVLink benchmark -- and collects the artifacts. topology.py renders the host's GPU / NVLink / NUMA / transport topology (text and Graphviz), plot.py renders the figures, and summary.py assembles a single summary.html. Each step degrades gracefully when an input or tool is absent. |
||
|---|---|---|
| .. | ||
| gpu_bench.py | ||
| plot.py | ||
| run_demo.sh | ||
| summary.py | ||
| topology.py | ||