How to perform a store operation on a part of a Tensor? about triton HOT 1 CLOSED

    num_blocks = tl.cdiv(N, BLOCK_SIZE)
    block_offs = tl.arange(0, NUM_BLOCKS)
    per_block_offs = tl.arange(0, BLOCK_SIZE)
    all_offs = block_offs[None, :] * BLOCK_SIZE + per_block_offs[:, None]
    
    # 2d pointers of shape [NUM_BLOCKS, BLOCK_SIZE]
    a = tl.load(a_ptr + all_offs,
                mask=all_offs < N and block_offs[None, :] < num_blocks,
                other=0.0)
    b = tl.load(b_ptr + all_offs,
                mask=all_offs  < N and block_offs[None, :] < num_blocks,
                other=0.0)
  
    # [NUM_BLOCKS, BLOCK_SIZE] --> NUM_BLOCKS * BLOCK_SIZE in which N elements will be filled others = 0
    accum_outs = tl.ravel(a * b)