julia> profile_all()
Method: mapreduce(f, op, A)
  for Array{Float64}:           246.070 milliseconds (15008 k allocations: 229 MB, 9.91% gc time)
  for NullableArray{Float64}:   417.401 milliseconds (15008 k allocations: 458 MB, 15.67% gc time)
  for DataArray{Float64}:       278.743 milliseconds (15008 k allocations: 229 MB, 9.67% gc time)
Method: reduce(op, A)
  for Array{Float64}:             2.629 milliseconds (1 allocation: 16 bytes)
  for NullableArray{Float64}:     5.286 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         2.851 milliseconds (2 allocations: 96 bytes)
Method: sum(A)
  for Array{Float64}:             2.810 milliseconds
  for NullableArray{Float64}:     6.432 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         5.209 milliseconds (2 allocations: 96 bytes)
Method: sum(f, A)
  for Array{Float64}:           252.220 milliseconds (15008 k allocations: 229 MB, 10.83% gc time)
  for NullableArray{Float64}:   424.200 milliseconds (15008 k allocations: 458 MB, 15.71% gc time)
  for DataArray{Float64}:       269.271 milliseconds (15008 k allocations: 229 MB, 9.69% gc time)
Method: prod(A)
  for Array{Float64}:             8.409 milliseconds
  for NullableArray{Float64}:    23.911 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         8.767 milliseconds (2 allocations: 96 bytes)
Method: prod(f, A)
  for Array{Float64}:           255.930 milliseconds (15008 k allocations: 229 MB, 10.65% gc time)
  for NullableArray{Float64}:       410 milliseconds (15008 k allocations: 458 MB, 16.58% gc time)
  for DataArray{Float64}:       272.405 milliseconds (15008 k allocations: 229 MB, 9.38% gc time)
Method: minimum(A)
  for Array{Float64}:             3.702 milliseconds
  for NullableArray{Float64}:    33.607 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         6.642 milliseconds (2 allocations: 96 bytes)
Method: minimum(f, A)
  for Array{Float64}:           238.662 milliseconds (10000 k allocations: 153 MB, 7.49% gc time)
  for NullableArray{Float64}:   517.422 milliseconds (15000 k allocations: 381 MB, 13.49% gc time)
  for DataArray{Float64}:       254.674 milliseconds (10000 k allocations: 153 MB, 7.48% gc time)
Method: maximum(A)
  for Array{Float64}:             7.461 milliseconds
  for NullableArray{Float64}:    24.961 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         6.270 milliseconds (2 allocations: 96 bytes)
Method: maximum(f, A)
  for Array{Float64}:           239.670 milliseconds (10000 k allocations: 153 MB, 8.19% gc time)
  for NullableArray{Float64}:   511.292 milliseconds (15000 k allocations: 381 MB, 14.05% gc time)
  for DataArray{Float64}:       245.280 milliseconds (10000 k allocations: 153 MB, 8.63% gc time)
Method: sumabs(A)
  for Array{Float64}:             2.756 milliseconds
  for NullableArray{Float64}:    28.615 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         2.654 milliseconds (2 allocations: 96 bytes)
Method: sumabs2(A)
  for Array{Float64}:             3.294 milliseconds
  for NullableArray{Float64}:    28.927 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         3.470 milliseconds (2 allocations: 96 bytes)

julia> profile_skip(true)
Comparison of skipnull/skipNA methods

f := IdFun(), op := AddFun()
mapreduce(f, op, X; skipnull/skipNA=true) (0 missing entries)
  for NullableArray{Float64}:     8.569 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         2.741 milliseconds (2 allocations: 96 bytes)

reduce(op, X; skipnull/skipNA=true) (0 missing entries)
  for NullableArray{Float64}:     9.563 milliseconds (3 allocations: 192 bytes)
  for DataArray{Float64}:         2.736 milliseconds (3 allocations: 176 bytes)

mapreduce(f, op, X; skipnull/skipNA=true) (~half missing entries)
  for NullableArray{Float64}:    52.518 milliseconds (2 allocations: 96 bytes)
  for DataArray{Float64}:        53.641 milliseconds (2 allocations: 96 bytes)

reduce(op, X; skipnull/skipNA=true) (~half missing entries)
  for NullableArray{Float64}:    54.764 milliseconds (3 allocations: 176 bytes)
  for DataArray{Float64}:        51.647 milliseconds (3 allocations: 176 bytes)

julia> profile_skip(false)
Comparison of skipnull/skipNA methods

f := IdFun(), op := AddFun()
mapreduce(f, op, X; skipnull/skipNA=false) (0 missing entries)
  for NullableArray{Float64}:     5.855 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:         2.911 milliseconds (2 allocations: 96 bytes)

reduce(op, X; skipnull/skipNA=false) (0 missing entries)
  for NullableArray{Float64}:     9.874 milliseconds (3 allocations: 192 bytes)
  for DataArray{Float64}:         5.123 milliseconds (3 allocations: 176 bytes)

mapreduce(f, op, X; skipnull/skipNA=false) (~half missing entries)
  for NullableArray{Float64}:     5.486 milliseconds (2 allocations: 112 bytes)
  for DataArray{Float64}:           753 nanoseconds  (1 allocation: 80 bytes)

reduce(op, X; skipnull/skipNA=false) (~half missing entries)
  for NullableArray{Float64}:     6.782 milliseconds (3 allocations: 192 bytes)
  for DataArray{Float64}:        50.259 milliseconds (3 allocations: 176 bytes)

julia> profile_skip_impl()
Comparison of internal skip methods:
mapreduce_impl_skipnull(f, op, X) (0 missing entries)
  for NullableArray{Float64}:     7.470 milliseconds (1 allocation: 16 bytes)
  for DataArray{Float64}:         3.644 milliseconds (1 allocation: 16 bytes)

mapreduce_impl_skipnull(f, op, X) (~half missing entries)
  for NullableArray{Float64}:    29.196 milliseconds (1 allocation: 16 bytes)
  for DataArray{Float64}:        51.622 milliseconds (1 allocation: 16 bytes)