Skip to main content

Mojo function

load_relaxed

load_relaxed[dtype: DType, //, *, scope: Scope = Scope(6), memory: Bool = True, alignment: Int = align_of[Scalar[dtype]]()](ptr: UnsafePointer[Scalar[dtype], address_space=address_space, mut=mut, origin=origin]) -> Scalar[dtype]

Performs an atomic load with relaxed memory ordering semantics.

On NVIDIA, maps to PTX ld.relaxed; on AMD, maps to POP atomic load with MONOTONIC ordering.

Returns:

Scalar

Was this page helpful?