Re: Global array operations: a performance hit?

On 18 Jun., 14:18, deltaquattro <deltaquat...@xxxxxxxxx> wrote:
On 18 Giu, 11:41, Dennis Wassel <dennis.was...@xxxxxxxxxxxxxx> wrote:

I'm not flaming and I don't think you are a novice: on the contrary,
since you talked about the possibility of you being involved in gcc
development, I think you know Fortran way better than me.
Anyway, I would never say "STFU n00b" to anyone under any
circumstances, if I understand what that means (I'm not sure since
because English is not my mother tongue, and I don't live in an
English speaking country).
The point is that when I posted a genuine doubt about speed of whole
array operations, you replied asking me if I have been clever enough
in doing my timings. When Richard confirmed my doubts, you were ready
to accept his words. To me, it seemed as if you were dismissing my
doubts just because I'm not an expert as Richard, robin, glenn, Paul,
Steve and all the many other guys on this newsgroup.

*sigh* I'm afraid it sounded exactly like that.

Sorry if I sounded as if I just dismissed your doubts as pointless, I
am in no position to do that; I wanted to point out that proper timing
is nontrivial, and whether you're aware of that. Apparently you are!
With Richard, his post (and google, I admit!) gives me some idea about
his background, so I'll tend to trust in what this guy says.

Besides, the fact that array operations may indeed be slower than
explicit loops was (and still is) hard for me to believe. Learned
something today!

If this isn't so, then I apologize to you and to the rest of the
newsgroup: I have misinterpreted your words, maybe also because of my
imperfect knowledge of English.

Best Regards,

deltaquattro

No offence meant, and none taken (I hope)!

BTT:
Is (compiler) optimisation of array operations that much of a
nontrivial task or just something that nobody has come around to
doing, yet? Judging from their mailing list, gfortran seems to have
many construction sites and a number of them apparently with higher
priorities.

But deltaquattro reported this against Compaq, and I also remember
someone from the gfortran mailing list mentioning that ifort shines on
DO-loop optimisation. Alas, I really feel like digging into this, but
got to get my current project done first.

Cheers,
Dennis
.

Relevant Pages

  • Re: 1001 Books You Must Trash Before You Die
    ... More to her point, I think, were any English accounts, anonymous pamphlets ... England during the time of Richard III? ... I may have to reread a good scholarly history of Richard III, ... their brother the king's approval, or Warwick's actions, or the ...
    (rec.arts.sf.written)
  • Re: morosely for morbidly
    ... Richard, Richard. ... This isn't the first time that native speakers of English have acted ... For weeks and months a number of morons insisted ...
    (alt.usage.english)
  • Re: RefersToRange.Parent gives back an error(from book written by
    ... formula (in English). ... >> The parent of a range is the worksheet (and the parent of the worksheet is the ... >> Richard wrote: ... >> Dave Peterson ...
    (microsoft.public.excel.programming)
  • Re: Chinese dictation
    ... > Richard> jotting down an English sentence, I cn omit sm vwls nd ... Both Chinese and English writings systems are ... >stroke shapes of Chinese characters, ...
    (sci.lang)