SpiceBench uses CLI flags for runtime behavior. See the CLI Reference for the complete flag list.
The current main spicebench binary exposes one built-in benchmark scenario:
| Scenario | Flag | Description |
|---|---|---|
| TPC-H | --scenario tpch |
Built-in TPC-H scenario and query workload |
The main binary does not currently expose separate --query-set, --scenario-query-file, or --query-overrides flags.
The --table-format flag declares the storage format for benchmark tables:
| Format | Value | Description |
|---|---|---|
| Parquet | parquet |
Apache Parquet (default) |
| Iceberg | iceberg |
Apache Iceberg |
| Delta | delta |
Delta Lake |
This value is passed to the system adapter in setup metadata for table creation in the appropriate format.
SpiceBench attaches metadata to each run for cross-system comparison and adapter setup:
| Field | CLI Flag / Source | Default / Behavior | Description |
|---|---|---|---|
table_format |
--table-format |
parquet |
Dataset table format |
executor_instance_type |
--executor-instance-type |
unknown |
Executor hardware class |
scenario |
--scenario |
required | Benchmark scenario |
system_adapter_name |
--system-adapter-name |
system_adapter |
Logical target-system identifier |
etl_bucket |
--etl-bucket |
spiceai-public-datasets |
Source data bucket |
etl_prefix |
--etl-prefix |
data-gen |
Source data prefix |
etl_version |
derived from --scale-factor |
format_scale_factor(scale_factor) |
Version path segment passed to adapters |
scale_factor |
--scale-factor |
1.0 |
Benchmark dataset scale factor |
scheduler_state_location |
--scheduler-state-location |
omitted unless provided | Optional shared scheduler-state location |
SpiceBench sends this metadata to the adapter in the setup request. A subset is also attached to exported metrics as resource attributes or per-metric attributes, depending on the metric pipeline.