Function dispatch_strided_copy_f32 

pub fn dispatch_strided_copy_f32(
    encoder: &mut CommandEncoder,
    registry: &mut KernelRegistry,
    device: &DeviceRef,
    src: &MlxBuffer,
    dst: &MlxBuffer,
    params: &StridedCopyParams,
) -> Result<()>

Dispatch a strided copy operation on the GPU.

Copies a 2D strided tensor to contiguous layout: dst[row * cols + col] = src[row * stride_row + col * stride_col]

Arguments

• encoder - Command encoder to record the dispatch into.
• registry - Kernel registry (must have strided_copy_f32 registered).
• device - Metal device for pipeline compilation.
• src - Source buffer (f32, strided layout).
• dst - Destination buffer (f32, contiguous output).
• params - Copy parameters (rows, cols, strides).

Errors

Returns MlxError::InvalidArgument if dimensions are 0 or buffers are too small.