HardwareOptions#

class nvmolkit.types.HardwareOptions(
preprocessingThreads: int = -1,
batchSize: int = -1,
batchesPerGpu: int = -1,
gpuIds: Iterable[int] | None = None,
)#

Configures GPU hardware settings for batch operations.

Use this class to control threading and batching behavior for GPU-accelerated workflows such as ETKDG embedding and MMFF optimization.

Parameters:

  • preprocessingThreads – Number of CPU threads for preprocessing. Use -1 to auto-detect and use all CPU threads.

  • batchSize – Number of conformers processed per batch. -1 for default.

  • batchesPerGpu – Number of batches processed concurrently on each GPU. Use -1 for default.

  • gpuIds – GPU device IDs to target. Provide an empty list to use all available GPUs.

property preprocessingThreads: int#

Number of CPU threads for preprocessing. -1 auto-detects.

property batchSize: int#

Number of conformers per batch. -1 selects an auto-tuned value.

property batchesPerGpu: int#

Batches processed concurrently on each GPU. -1 auto.

property gpuIds: List[int]#

GPU device IDs to target. Empty list uses all available GPUs.

to_dict() → dict[str, Any]#

Return a JSON-serializable dictionary of this object’s fields.

The returned dictionary can be persisted with json.dump() and round-tripped through from_dict().

classmethod from_dict(
data: dict[str, Any],
) → HardwareOptions#

Create a HardwareOptions from a dictionary produced by to_dict().

Unknown keys are rejected so callers catch typos early. Missing keys fall back to the constructor defaults.