Re: Matrix Multiplication

On 2008-01-05 08:53:36 -0400, Evgenii Rudnyi <usenet@xxxxxxxxx> said:

On Jan 5, 11:06 am, b52b <jam...@xxxxx> wrote:
On Jan 5, 8:37 am, Evgenii Rudnyi <use...@xxxxxxxxx> wrote:
. I just wonder why Gordon prefers to keep silent when it> comes to the code.

This is matter of ethics. Writing an educational text is hard because
you should be up to date. That's what Gordon said in my opinion.
You can't blame him - it is your article.

Sorry, but I would disagree. My text is completely up to date. The
current practice for linear algebra with dense matrices is to use
LAPACK (by the way, it is written in Fortran 77) together with the
optimized BLAS. From what a programming language to call LAPACK and
optimized BLAS does not matter.

Also if you read the conclusion, it says

http://matrixprogramming.com/MatrixMultiply/
-------------------------------
I hope that this chapter has convinced you that it is very important
to use libraries even for seemingly simple things, like matrix
multiplication.
--------------------------------

I am pretty sure it concerns the newer Fortran as well. There are no
miracles, either one uses optimized DGEMM or performance suffers. The
use of optimized DGEMM can be behind the scene, as in NumPy, Matlab
and Mathematica, this does not matter.

My advice : sent your request to comp.lang.fortran
Otherwise this is another face of the language war.

This what I wanted to escape, as personally I believe that good
performance can be achieved in any programming language provided that
one understands the technology.

My text is more about technology, not about a programming language.

Evgenii

Thanks to Dave Dobson for a sensible response.

It is your article to write. You have picked a hard topic and asked for
comments. You got them. They pointed out consequences of your choice to
switch to C++ in 1993 and to extol its virtues. Your choice of systems
to compare is certainly idiosyncratic. Python, C++ vrs C and F77 seems
to be more a setup for language wars than about Matrix Programming.
Insisting that comments come with "your code" is following the best
newsgroup language war traditions.

If you want to write about matrix programming then where is the discussion
of MatLab and its clones? Fortran 90 lives well in that crowd with its
array operations. Matrix multiply is one of the supplied intrinsics. To
ignore that reality of current practice is to either be misinformed or
out of date. Once one has gotten the concepts sorted out with vector and
matrix capabilities then one can go back to the element based operations.
To dismiss F90 with the comment that you only know F77 would appear
to be an admission that you do not want to follow the contemporary style of
use of the array notations. Comments on your pedogogy seem to be treated
as comments on your abilities. Such defensiveness is an overreation but
as the saying goes "If the shoe fits ...".




.

Relevant Pages

  • Re: Matrix Multiplication
    ... I just wonder why Gordon prefers to keep silent when it> comes to the code. ... This is matter of ethics. ... From what a programming language to call LAPACK and ... optimized BLAS does not matter. ...
    (sci.math.num-analysis)
  • Security: Unix versus Windows
    ... We all know that most operating systems and software suffer from identical ... flaws no matter what their origin and dev model is. ... Writing everything in Java [a programming language created by Sun] ...
    (SecProg)
  • Re: New symbolic/numeric/dynamic/intuitive programming language
    ...  So does this work in Flaming Thunder? ... Just a matter of personal taste, ... ambiguities is hardly a feature I desire in a programming language (or ...
    (sci.math)
  • Re: Usage of 8-byte integers (Was Re: Usage of complex numbers?)
    ... >> If X and Y are positive integers then A ultimately diverges no matter ... >> programming language or precision you care to use. ...
    (comp.lang.fortran)