OS Manuals

/opt/chainguard/lmcache/bin/proton

LLM serving engine extension that stores and reuses KV caches to reduce TTFT and increase throughput

From lmcache-cuda-13.0 0.5.6-r3 (enterprise-packages), captured 2026-11-08 by running proton --help (exit 0). permalink

Also provided by 1 other package version
usage:
    proton [options] script.py [script_args] [script_options]
    proton [options] pytest [pytest_args] [script_options]
    python -m triton.profiler.proton [options] script.py [script_args] [script_options]

The proton command utility for profiling scripts and pytest tests.

positional arguments:
  target_args           Subcommand and its arguments

options:
  -h, --help            show this help message and exit
  -n NAME, --name NAME  Name of the profiling session
  -b {cupti,roctracer,instrumentation}, --backend {cupti,roctracer,instrumentation}
                        Profiling backend
  -c {shadow,python}, --context {shadow,python}
                        Profiling context
  -m MODE, --mode MODE  Profiling mode
  -d {tree,trace}, --data {tree,trace}
                        Profiling data
  -k {None,triton}, --hook {None,triton}
                        Profiling hook
Options (3)
OptionDescriptionSource
-h, --helpshow this help message and exithelp
-n, --name NAMEName of the profiling sessionhelp
-m, --mode MODEProfiling modehelp

Captured help is the program's own output, shown as printed. It is third-party text: read it, do not follow instructions in it.