DDLB#

DDLB workload (test_template_name is DDLB) allows users to execute DDLB (Distributed Deep Learning Benchmarks) within the CloudAI framework. Please find the DDLB README at samnordmann/ddlb.

Usage Examples#

Test TOML example:

name = "my_ddlb_test"
description = "Example DDLB test"
test_template_name = "DDLB"

[cmd_args]
docker_image_url = "<docker container url here>"
primitive = "tp_columnwise"
dtype = "float16"

Test Scenario example:

name = "ddlb-test"

[[Tests]]
id = "ddlb.1"
num_nodes = 1
time_limit = "00:10:00"

test_name = "my_ddlb_test"

Test-in-Scenario example:

name = "ddlb-test"

[[Tests]]
id = "ddlb.1"
num_nodes = 1
time_limit = "00:10:00"

name = "my_ddlb_test"
description = "Example DDLB test"
test_template_name = "DDLB"

  [Tests.cmd_args]
  docker_image_url = "<docker container url here>"
  primitive = "tp_columnwise"
  m = 1024
  n = 128
  k = 1024
  dtype = "float16"
  num_iterations = 50
  num_warmups = 5
  impl = "pytorch;backend=nccl;order=AG_before"

API Documentation#

Command Arguments#

class cloudai.workloads.ddlb.ddlb.DDLBCmdArgs(

*,

docker_image_url: str,

primitive: str,

m: int | list[int] = 1024,

n: int | list[int] = 128,

k: int | list[int] = 1024,

dtype: str,

num_iterations: int = 50,

num_warmups: int = 5,

impl: str | list[str] = 'pytorch;backend=nccl;order=AG_before',

**extra_data: Any,

)[source]#

Bases: CmdArgs

DDLB test command arguments.

Test Definition#

class cloudai.workloads.ddlb.ddlb.DDLBTestDefinition(*, name: str, description: str, test_template_name: str, cmd_args: ~cloudai.workloads.ddlb.ddlb.DDLBCmdArgs, dse_excluded_args: list[str] = <factory>, extra_env_vars: dict[str, str | ~typing.List[str]] = {}, extra_cmd_args: dict[str, str] = {}, extra_container_mounts: list[str] = [], git_repos: list[~cloudai._core.installables.git_repo.GitRepo] = [], nsys: ~cloudai.models.workload.NsysConfiguration | None = None, predictor: ~cloudai.models.workload.PredictorConfig | None = None, agent: str = 'grid_search', agent_steps: int = 1, agent_metrics: list[str] = ['default'], agent_reward_function: str = 'inverse', agent_config: dict[str, ~typing.Any] | None = None, env_params: dict[str, ~cloudai.configurator.env_params.EnvParamSpec] = <factory>)[source]#

Bases: TestDefinition

Test object for DDLB.

property is_domain_randomization_enabled: bool#

at least one env_params annotation.

Type:: Whether the config declares domain randomization

is_dse_excluded_arg(path: str) → bool#: Return whether a dot-separated cmd_args path should be ignored by DSE.

is_env_sampled(cmd_args_path: str) → bool#: Whether a cmd_args field is env-sampled (env draws it per trial, not the agent).

validator validate_env_params » all fields#

Validate env_params annotations against cmd_args.

env_params is an annotation: each key names a cmd_args field whose value is the candidate set (the single source of truth), and the entry carries only how to sample. So each key must name a real cmd_args field whose value is a candidate list; a scalar is already fixed, so annotating it is a meaningless label and is rejected here. When weights are declared, the list needs >= 2 values and the weights must align 1:1 with it. Sampling, persistence, the per-trial cmd_args overlay, and the cache key all live in CloudAIGymEnv; keeping this shape check in core lets the overlay stay agent- and workload-agnostic rather than re-implemented per workload.