Skip to content

Runner: make MPI bind-to configurable #797

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
alephpiece wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Runner: make MPI bind-to configurable #797

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
29c250b
Runner: make MPI bind-to configurable
alephpiece Mar 27, 2026
a2e59d1
Runner: update docs
alephpiece Mar 27, 2026
a440d09
Merge branch 'main' into one/runner-bind-to-configurable
alephpiece Mar 27, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions docs/superbench-config.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -370,6 +370,7 @@ proc_num: int
node_num: int
env: dict
mca: dict
bind_to: string
prefix: str
parallel: bool
```
Expand Down Expand Up @@ -403,6 +404,7 @@ Some attributes may only be suitable for specific mode.
| `prefix` | ✓ | ✘ | ✘ |
| `env` | ✓ | ✓ | ✓ |
| `mca` | ✘ | ✘ | ✓ |
| `bind_to` | ✘ | ✘ | ✓ |
| `parallel` | ✓ | ✘ | ✘ |
| `pattern` | ✘ | ✘ | ✓ |

Expand Down Expand Up @@ -452,6 +454,16 @@ MCA (Modular Component Architecture) frameworks, components, or modules to use i
in a flatten key-value dictionary.
Only available for `mpi` mode.

### `bind_to`

Process binding policy passed to `mpirun -bind-to`.
Only available for `mpi` mode.

Use this option when a benchmark needs to override the runner's default MPI binding behavior,
for example when the benchmark implements its own topology-aware CPU/NUMA affinity logic.

* default value: `numa`

### `parallel`

Whether run benchmarks in parallel (all ranks at the same time) or in sequence (one rank at a time).
Expand Down
5 changes: 4 additions & 1 deletion superbench/runner/runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,8 @@ def __validate_sb_config(self): # noqa: C901
'btl_tcp_if_exclude': 'lo,docker0',
'coll_hcoll_enable': 0,
}
if 'bind_to' not in mode:
self._sb_benchmarks[name].modes[idx].bind_to = 'numa'
for key in ['PATH', 'LD_LIBRARY_PATH', 'SB_MICRO_PATH', 'SB_WORKSPACE']:
self._sb_benchmarks[name].modes[idx].env.setdefault(key, None)
if 'pattern' in mode:
Expand Down Expand Up @@ -182,13 +184,14 @@ def __get_mode_command(self, benchmark_name, mode, timeout=None):
'-tag-output ' # tag mpi output with [jobid,rank]<stdout/stderr> prefix
'-allow-run-as-root ' # allow mpirun to run when executed by root user
'{host_list} ' # use prepared hostfile or specify nodes and launch {proc_num} processes on each node
'-bind-to numa ' # bind processes to numa
'-bind-to {bind_to} ' # bind processes according to mode config
'{mca_list} {env_list} {command}'
).format(
trace=trace_command,
host_list=f'-host localhost:{mode.proc_num}' if 'node_num' in mode and mode.node_num == 1 else
f'-hostfile hostfile -map-by ppr:{mode.proc_num}:node' if 'host_list' not in mode else '-host ' +
','.join(f'{host}:{mode.proc_num}' for host in mode.host_list),
bind_to=mode.bind_to,
mca_list=' '.join(f'-mca {k} {v}' for k, v in mode.mca.items()),
env_list=' '.join(
f'-x {k}={str(v).format(proc_rank=mode.proc_rank, proc_num=mode.proc_num)}'
Expand Down