The signature for af_add:
af_add (af_array *out, const af_array lhs, const af_array rhs, const bool batch)
with lhs and rhs as parameter names, I don't think J users would get lost if
they called the "cover function" dyadically the same way they call +
I'm not suggesting altering any existing code that does follow the "C spec"
closely. Just adding adverbs on top of them to provide an intuitive J
interface to the functions.
consider the J function
insert =: 1 : 'u/"x y' (_&$: :)
it is "monadic only" version of /. With optional rank parameter as dyadic
number
+ insert i.3 3
9 12 15
1 + insert i.3 3
3 12 21
J does not need such an adverb ("x is easy to type/add and "name context
free"), but "lesser languages" do. Although when they do use such an
implementation, they capture the full flexibility of ranked-modified insert. A
more important flexibility than being able to turn insert into, a very
different adverb, table.
this insert adverb definition would work like a lesser language modifier....
like af_sum, af_product group.
One of the reasons to make this consistency change is to simplify conversion to
and from J<->arrayfire including the great matmul demonstration you have
provided.
On Wednesday, January 26, 2022, 12:23:52 p.m. EST, Eric Iverson
<eric.b.iver...@gmail.com> wrote:
Pascal,
Interesting ideas on arrayfire args.
I will probably resist the idea of making dop verbs dyadic. I think it
better to stick closer to the arrayfire syntax.
But it makes sense to avoid unnecessary boxing.
af_add_jaf_ afad;afbd NB. works
af_add_jaf_ afad,afbd NB. fails, but probably should work
On Wed, Jan 26, 2022 at 10:03 AM 'Pascal Jasmin' via Programming <
programm...@jsoftware.com> wrote:
> An interface suggestion would be to transform all verbs that currently
> take 2 boxed arguments into unboxed dyad arguments.
>
> this can be done by adding the adverb (@:;) for the dop class verbs. and
> (@:;~) for the reduce class verbs such that the operating axis is the x
> term. The reduce class could have the additional adverb (0&$: : ) to make
> supplying the axis an optional parameter. The explicitly constructed
> functions in arrayfire.ijs that take 2 boxed y parameters (matmul) can also
> be decorated with (@: ;)
>
> These changes would more naturally leverage J's processing/modifier
> capabilities, and processing list of arrayfire array handles. Arrayfire
> sparse support seems to be limited to rank 2 arrays. J could circumvent
> these limitations by shaping multiple arrayfire handles managing the
> nesting logic in J instead.
>
> There are many uses of arrayfire that are made easier with a list of array
> handles.
>
>
>
>
> On Tuesday, January 25, 2022, 04:39:36 p.m. EST, Eric Iverson <
> eric.b.iver...@gmail.com> wrote:
>
>
>
>
>
> Pascal,
>
> Thanks for the feedback. Coming from you, it means a lot to me.
>
> There are some rough edges for windows and mac (I work mostly in linux)
> that I will fix tomorrow morning.
>
> On Tue, Jan 25, 2022 at 3:52 PM 'Pascal Jasmin' via Programming <
> programm...@jsoftware.com> wrote:
>
> > looks good. A small bug:
> >
> > if you define
> >
> > reduce_z_ =: ]: NB. adverb
> >
> > prior to loading arrayfire.ijs
> >
> > load will fail.
> >
> > The reason is the families.ijs file creates a bunch of forks with their
> > middle verbs defined after the forks are created.
> >
> > An easy fix is to define all of the explicit verbs used in families.ijs
> > above the line that "require families.ijs"
> >
> >
> > fixed it for me.
> >
> >
> >
> > On Tuesday, January 25, 2022, 10:56:54 a.m. EST, Eric Iverson <
> > eric.b.iver...@gmail.com> wrote:
> >
> >
> >
> >
> >
> > ~addons/math/arrayfire is a new addon for J903 and later.
> >
> > The addon allows high performance computing of matrix product
> > and many other mathematical array operations. Multiple cpu
> > cores and hardware such as Nvidia gpus are used as appropriate.
> >
> > Browse https://arrayfire.org for more info.
> >
> > The addon will prompt you for the ArrayFire software install if
> > it is not already installed.
> >
> > Use pacman to install the addon. For example,
> > load'pacman'
> > 'install' jpkg 'math/arrayfire'
> >
> > Get started:
> > load '~addons/math/arrayfire/arrayfire.ijs'
> > man_jaf_ 'intro'
> > ...
> > init_jaf_ 'cpu'
> > tut_jaf_ 'basic' NB. run basic tutorial
> > ...
> > man_jaf_ 'credits' NB. credit to some J-ArrayFire pioneers
> > ...
> >
> > *** preliminary version
> > We think this is an important step for J in high performance computing.
> > If this proves to be the case, Jsoftware will actively pursue further
> > development in this area.
> >
> > *** from the addon description
> > cd bindings and utilities for ArrayFire shared library
> >
> > https://arrayfire.org
> >
> > "ArrayFire is a high performance software library for
> > parallel computing with an easy-to-use API. Its array based
> > function set makes parallel programming more accessible."
> >
> > J-ArrayFire cd bindings make it easy to have
> > high performance computing for things like matrix multiply
> > with cpu cores, Nvidia gpus, and opencl.
> > ***
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
>
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
