For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).
Mojo struct
ProduceContext
struct ProduceContext[pipeline_origin: MutOrigin, num_stages: Int, Backend: PipelineBackend = NvidiaMbarBackend[num_stages]]
Context manager for producing one pipeline stage.
- enter: Waits for consumer to be ready, returns ref to stage
- exit: Releases the stage (advances producer)
Note: The actual production signal (mma_arrive) is kernel-specific and must be called by the user before exiting the context.
Parameters
- pipeline_origin (
MutOrigin): Origin of the pipeline reference. - num_stages (
Int): Number of pipeline stages. - Backend (
PipelineBackend): Pipeline synchronization backend (defaults toNvidiaMbarBackend).
Fields
- pipeline (
Pointer[ProducerConsumerPipeline[num_stages, Backend], pipeline_origin, _safe=True]):
Implemented traits
Methods
__init__
def __init__(out self, pipeline: Pointer[ProducerConsumerPipeline[num_stages, Backend], pipeline_origin, _safe=True])
__del__
def __del__(deinit self)
__enter__
def __enter__(mut self) -> ref[self._stage._value] ProducerStage[pipeline_origin, num_stages, Backend]
Wait for consumer and return reference to stage.
Returns:
ref[self._stage._value] ProducerStage[pipeline_origin, num_stages, Backend]
__exit__
def __exit__(mut self)
Release the stage (advances producer).
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!