This is one of my pet peeves. "Random Access" memory is far from random when you look at the time it takes to do the accesses. With modern memories, accessing a column can be 20 to 40x slower than accessing a row. This is particularly irritating when doing AI training, where training reuses 4-d tensors transposed, a very painful operation.
In FORTRAN days, I once used a vector package in which you described a vector by giving the first element, the second element, and a count. So you could describe rows, columns, a matrix diagonal, and even rows and columns from back to front. Fortran passed arguments by address, which made the whole thing easy and fast.
Steve
----- Original Message -----
From:
"Doug McIlroy" <doug@cs.dartmouth.edu>
To:
<tuhs@tuhs.org>, <jnc@mercury.lcs.mit.edu>
Cc:
Sent:
Tue, 17 Sep 2019 13:31:52 -0400
Subject:
Re: [TUHS] block operations in editors, was My EuroBSDcon talk
Noel Chiappa wrote:
> > From: Doug McIlroy
>
> > the absence of multidemensional arrays in C.
>
>?? From the 'C Reference Manual' (no date, but circa 'Typesetter C'), pg. 11:
>
> "If the unadorned declarator D would specify an n-dimensional array .. then
> the declarator D[in+1] yields an (n+1)-dimensional array"
>
>
>I'm not sure if I've _ever_ used one, but they are there.
Yes, C allows arrays of arrays, and I've used them aplenty.
However an "n-dimensional array" has one favored dimension,
out of which you can slice an array of lower dimension. For
example, you can pass a row of a 2D array to a function of a
1D variable, but you can't pass a column. That asymmetry
underlies my assertion. In Python, by contrast, n-dimensional
arrays can be sliced on any dimension.
Doug