Mojo function
naive_reduce_kernel
naive_reduce_kernel[type: DType](dst_buf: UnsafePointer[SIMD[type, 1]], src_buf: UnsafePointer[SIMD[type, 1]], num_elements: Int)
A simple reduction kernel that adds source buffer values to destination buffer.
Arguments: dst_buf: Destination buffer to accumulate results src_buf: Source buffer containing values to add num_elements: Number of elements to process
Each thread handles multiple elements with striding for coalesced memory access.
Parameters:
- type (
DType): DType - The data type of the values being reduced.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!