NIXL CTPerf
This workload (test_template_name is NixlPerftest) runs NIXL performance testing suite for comprehensive network performance evaluation.
Test TOML example:
name = "my_nixl_perftest_test"
description = "Example NIXL Perftest test"
test_template_name = "NixlPerftest"
[cmd_args]
docker_image_url = "<docker container url here>"
subtest = "sequential-ct-perftest"
num_user_requests = 1
batch_size = 1
num_prefill_nodes = 1
num_decode_nodes = 1
prefill_tp = 4
decode_tp = 4
isl_mean = 10000
isl_scale = 3000
model = "deepseek-r1-distill-llama-70b"
Test Scenario example:
name = "nixl-perftest-test"
[[Tests]]
id = "perftest.1"
num_nodes = 2
time_limit = "00:20:00"
test_name = "my_nixl_perftest_test"
Test-in-Scenario example:
name = "nixl-perftest-test"
[[Tests]]
id = "perftest.1"
num_nodes = 2
time_limit = "00:20:00"
name = "my_nixl_perftest_test"
description = "Example NIXL Perftest test"
test_template_name = "NixlPerftest"
[Tests.cmd_args]
docker_image_url = "<docker container url here>"
subtest = "sequential-ct-perftest"
num_user_requests = 100
batch_size = 1
num_prefill_nodes = 1
num_decode_nodes = 1
prefill_tp = 8
decode_tp = 8
model = "deepseek-r1-distill-llama-70b"
[Tests.extra_env_vars]
CUDA_VISIBLE_DEVICES = "$SLURM_LOCALID"
Command Arguments
- class cloudai.workloads.nixl_perftest.nixl_perftest.NixlPerftestCmdArgs(*, docker_image_url: str, subtest: ~typing.Literal['sequential-ct-perftest'], perftest_script: str = '/workspace/nixl/benchmark/kvbench/main.py', matgen_script: str = '/workspace/nixl/benchmark/kvbench/test/inference_workload_matgen.py', python_executable: str = 'python', etcd_path: str = 'etcd', wait_etcd_for: int = 60, num_user_requests: int | list[int], batch_size: int | list[int], num_prefill_nodes: int | list[int], num_decode_nodes: int | list[int], isl_mean: int | list[int] | None = None, isl_scale: int | list[int] | None = None, prefill_tp: int | list[int] = 1, prefill_pp: int | list[int] = 1, prefill_cp: int | list[int] = 1, decode_tp: int | list[int] = 1, decode_pp: int | list[int] = 1, decode_cp: int | list[int] = 1, model: str | list[str] | None = None, hidden_size: int | None = None, num_layers: int | None = None, num_heads: int | None = None, num_kv_heads: int | None = None, dtype_size: int | None = None, matgen_args: ~cloudai.workloads.nixl_perftest.nixl_perftest.MatgenCmdArgs = <factory>, **extra_data: ~typing.Any)[source]
Bases:
CmdArgsCmdArgs for NixlPerftestTestDefinition.
- docker_image_url: str
- subtest: Literal['sequential-ct-perftest']
- perftest_script: str
- matgen_script: str
- python_executable: str
- etcd_path: str
- wait_etcd_for: int
- num_user_requests: int | list[int]
- batch_size: int | list[int]
- num_prefill_nodes: int | list[int]
- num_decode_nodes: int | list[int]
- isl_mean: int | list[int] | None
- isl_scale: int | list[int] | None
- prefill_tp: int | list[int]
- prefill_pp: int | list[int]
- prefill_cp: int | list[int]
- decode_tp: int | list[int]
- decode_pp: int | list[int]
- decode_cp: int | list[int]
- model: str | list[str] | None
- num_layers: int | None
- num_heads: int | None
- num_kv_heads: int | None
- dtype_size: int | None
- matgen_args: MatgenCmdArgs
Test Definition
- class cloudai.workloads.nixl_perftest.nixl_perftest.NixlPerftestTestDefinition(*, name: str, description: str, test_template_name: str, cmd_args: NixlPerftestCmdArgs, extra_env_vars: dict[str, str | List[str]] = {}, extra_cmd_args: dict[str, str] = {}, extra_container_mounts: list[str] = [], git_repos: list[GitRepo] = [], nsys: NsysConfiguration | None = None, predictor: PredictorConfig | None = None, agent: str = 'grid_search', agent_steps: int = 1, agent_metrics: list[str] = ['default'], agent_reward_function: str = 'inverse')[source]
Bases:
TestDefinitionTestDefinition for NixlPerftest.
- cmd_args: NixlPerftestCmdArgs
- property docker_image: DockerImage
- property installables: list[Installable]
- constraint_check(tr: TestRun) → bool[source]