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
LaunchHostFunc
struct LaunchHostFunc
Enqueues a pre-packed host callback on the device's default stream.
Corresponds to CUDA's cuLaunchHostFunc. Accepts a 1-D int64 buffer of
shape [2] whose elements are raw pointer-sized integers:
payload[0]: address of avoid (*)(void *)trampoline function.payload[1]: address of an opaque user-data block owned by the trampoline (freed after the callback runs).
Both values are produced by max._core.driver._pack_host_func(fn) on
the Python side. Currently only CUDA streams support host callbacks;
non-CUDA backends raise at runtime.
Implemented traits
Methods
execute
static def execute[target: StringSlice[ImmStaticOrigin]](payload: ManagedTensorSlice[IOSpec[_, _].MutableInput, static_spec=payload.static_spec], ctx: DeviceContext)
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!