sub _do_dimension ( $m, $w ) {
    $m -> slice([ 0, $w - 1 ], [])
       -> sumover
       -> slice( '*' )
       -> append( 
            $w >= ( $m-> dims )[ 0 ]
                ? empty
                : $m-> slice([ $w, -1 ], []) - 
                  $m-> slice([ 0, -1 - $w ], []))
       -> cumusumover
       -> xchg( 0, 1 )
}

sub sms_WxH_PDL_sliding ( $m, $w, $h ) {
    _do_dimension 
        _do_dimension( $m, $w ), $h 
}

$m = 1 + sequence 4, 3;
print $m,
    sms_WxH_PDL_sliding( $m, 2, 2 ),
    sms_WxH_PDL_sliding( $m, 3, 3 ),
    sms_WxH_PDL_sliding( $m, 4, 3 ),
    sms_WxH_PDL_sliding( $m, 1, 1 );

__END__

# same (correct) output as above, skipped

+
    Time (s) vs. N (NxN submatrix, PDL: Double D [1500,1500] matrix)
      +-----------------------------------------------------------+
      |    +      +      +      +       +      +      +      +    |
      |                             A                             |
      |                         A                                 |
      |                      A          A                         |
  1.5 |-+                                  A                    +-|
      |                  A                                        |
      |                                                           |
      |                                        A                  |
      |                                                           |
      |               A                                           |
      |                                           A               |
    1 |-+                                                       +-|
      |                                                           |
      |                                               A           |
      |           A                                               |
      |                                                           |
      |                                                           |
      |                                                  A        |
  0.5 |-+                                                       +-|
      |                                                           |
      |        A                                                  |
      |                                                      A    |
      |                                                           |
      |    B          B                                           |
      |    BB  B  B      B   B  B   B   B  B          B           |
    0 |-+                                      B  B      B   B  +-|
      |    +      +      +      +       +      +      +      +    |
      +-----------------------------------------------------------+
           0     200    400    600     800    1000   1200   1400
                         sms_WxH_PDL_lags    A
                      sms_WxH_PDL_sliding    B
  +------+-------+-------+
  | N    | A     | B     |
  +------+-------+-------+
  | 2    | 0.025 | 0.100 |
  | 4    | 0.034 | 0.075 |
  | 6    | 0.016 | 0.084 |
  | 10   | 0.034 | 0.091 |
  | 20   | 0.028 | 0.084 |
  | 100  | 0.369 | 0.081 |
  | 200  | 0.800 | 0.078 |
  | 300  | 1.122 | 0.116 |
  | 400  | 1.416 | 0.050 |
  | 500  | 1.572 | 0.053 |
  | 600  | 1.644 | 0.022 |
  | 700  | 1.675 | 0.025 |
  | 800  | 1.587 | 0.019 |
  | 900  | 1.456 | 0.019 |
  | 1000 | 1.297 | 0.009 |
  | 1100 | 1.047 | 0.009 |
  | 1200 | 0.813 | 0.019 |
  | 1300 | 0.566 | 0.006 |
  | 1400 | 0.275 | 0.003 |
  +------+-------+-------+