warp.tile_max

warp.tile_max(

a: Tile[Scalar, tuple[int, ...]],

) → Tile[Scalar, tuple[1]]

Kernel

Cooperatively compute the maximum of the tile elements using all threads in the block.

param a:

The tile to compute the maximum from

returns:

A single-element tile holding the maximum value

Example:

@wp.kernel
def compute():

    t = wp.tile_arange(64, 128)
    s = wp.tile_max(t)

    print(s)

wp.launch_tiled(compute, dim=[1], inputs=[], block_dim=64)

Prints:

[127] = tile(shape=(1), storage=register)