Skip to navigation

NCCL

View as Markdown

This workload (test_template_name is NcclTest) allows users to execute NCCL benchmarks within the CloudAI framework.

Usage Example

Test TOML example:

name = "my_nccl_test"
description = "Example NCCL test"
test_template_name = "NcclTest"
[cmd_args]
docker_image_url = "nvcr.io#nvidia/pytorch:25.06-py3"

Test Scenario example:

name = "nccl-test"
[[Tests]]
id = "nccl.1"
num_nodes = 1
time_limit = "00:05:00"
test_name = "my_nccl_test"

Test-in-Scenario example:

name = "nccl-test"
[[Tests]]
id = "nccl.1"
num_nodes = 1
time_limit = "00:05:00"
name = "my_nccl_test"
description = "Example NCCL test"
test_template_name = "NcclTest"
[Tests.cmd_args]
docker_image_url = "nvcr.io#nvidia/pytorch:25.06-py3"
subtest_name = "all_reduce_perf_mpi"
iters = 100

API Documentation

Command Arguments

NCCLCmdArgs

NCCL test command arguments.

FieldTypeDefaultDescription
docker_image_urlstrRequired
subtest_nameLiteral['all_reduce_perf_mpi', 'all_gather_perf_mpi', 'alltoall_perf_mpi', 'alltoallv_perf_mpi', 'broadcast_perf_mpi', 'gather_perf_mpi', 'hypercube_perf_mpi', 'reduce_perf_mpi', 'reduce_scatter_perf_mpi', 'scatter_perf_mpi', 'sendrecv_perf_mpi', 'bisection_perf_mpi', 'all_reduce_perf', 'all_gather_perf', 'alltoall_perf', 'alltoallv_perf', 'broadcast_perf', 'gather_perf', 'hypercube_perf', 'reduce_perf', 'reduce_scatter_perf', 'scatter_perf', 'sendrecv_perf', 'bisection_perf'] | list[Literal['all_reduce_perf_mpi', 'all_gather_perf_mpi', 'alltoall_perf_mpi', 'alltoallv_perf_mpi', 'broadcast_perf_mpi', 'gather_perf_mpi', 'hypercube_perf_mpi', 'reduce_perf_mpi', 'reduce_scatter_perf_mpi', 'scatter_perf_mpi', 'sendrecv_perf_mpi', 'bisection_perf_mpi', 'all_reduce_perf', 'all_gather_perf', 'alltoall_perf', 'alltoallv_perf', 'broadcast_perf', 'gather_perf', 'hypercube_perf', 'reduce_perf', 'reduce_scatter_perf', 'scatter_perf', 'sendrecv_perf', 'bisection_perf']]'all_reduce_perf_mpi'
nthreadsint | list[int]1
ngpusint | list[int]1
minbytesstr | list[str]'32M'
maxbytesstr | list[str]'32M'
stepbytesstr | list[str]'1M'
opLiteral['sum', 'prod', 'min', 'max', 'avg', 'all'] | list[Literal['sum', 'prod', 'min', 'max', 'avg', 'all']]'sum'
datatypeLiteral['uint8', 'float'] | list[Literal['uint8', 'float']]'float'
rootint | list[int]0
itersint | list[int]20
warmup_itersint | list[int]5
agg_itersint | list[int]1
averageint | list[int]1
parallel_initint | list[int]0
checkint | list[int]1
blockingint | list[int]0
cudagraphint | list[int]0
stepfactorint | list[int] | NoneNone
use_deepep_matrixboolFalse
alltoallv_matrix_container_pathstr'/tmp/traffic_matrix.txt'
min_busbwfloat | NoneNone

Test Definition

NCCLTestDefinition

Test object for NCCL.

FieldTypeDefaultDescription
namestrRequired
descriptionstrRequired
test_template_namestrRequired
cmd_argscloudai.workloads.nccl_test.nccl.NCCLCmdArgsRequired
dse_excluded_argslist[str]Factory: list
extra_env_varsdict[str, str | typing.List[str]]{}
extra_cmd_argsdict[str, str]{}
extra_container_mountslist[str][]
git_reposlist[cloudai._core.installables.git_repo.GitRepo][]
nsyscloudai.models.workload.NsysConfiguration | NoneNone
predictorcloudai.models.workload.PredictorConfig | NoneNone
training_reportcloudai.models.workload.TrainingReportConfig | NoneNone
agentstr'grid_search'
agent_stepsint1
agent_metricslist[str]['default']
agent_reward_functionstr'inverse'
agent_configdict[str, Any] | NoneNoneAgent configuration.
env_paramsdict[str, cloudai.configurator.env_params.EnvParamSpec]Factory: dictEnvironment parameters sampled by the env per trial. Sibling to cmd_args; not part of the agent’s action space. CloudAIGymEnv samples, persists them in the trajectory output, and includes them in the trajectory cache key.