Re: B dny epsilon - this is not a bug, this is ? - Grupy dyskusyjne w eGospodarka.pl

eGospodarka.pl › Grupy › pl.comp.programming › Błędny epsilon - this is not a bug, this is ? › Re: B dny epsilon - this is not a bug, this is ?

Received: by 10.236.116.67 with SMTP id f43mr33302yhh.13.1352205151763; Tue, 06 Nov
2012 04:32:31 -0800 (PST)
Received: by 10.236.116.67 with SMTP id f43mr33302yhh.13.1352205151763; Tue, 06 Nov
2012 04:32:31 -0800 (PST)
Path: news-archive.icm.edu.pl!news.icm.edu.pl!plix.pl!newsfeed2.plix.pl!feed.xsnews.n
l!border-1.ams.xsnews.nl!ecngs!feeder.ecngs.de!nx01.iad01.newshosting.com!newsh
osting.com!69.16.185.16.MISMATCH!npeer02.iad.highwinds-media.com!news.highwinds
-media.com!feed-me.highwinds-media.com!c7no8594004qap.0!news-out.google.com!gf5
ni18188417qab.0!nntp.google.com!c7no8594003qap.0!postnews.google.com!glegroupsg
2000goo.googlegroups.com!not-for-mail
Newsgroups: pl.comp.programming
Date: Tue, 6 Nov 2012 04:32:31 -0800 (PST)
In-Reply-To: <k7ar3r$5oh$1@mx1.internetia.pl>
Complaints-To: g...@g...com
Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=193.189.80.39;
posting-account=EexxQQoAAAAkOfWz0VZRKLcHNpXJZLB9
NNTP-Posting-Host: 193.189.80.39
References: <50924bb1$0$1308$65785112@news.neostrada.pl>
<a...@g...com>
<50926d86$0$1216$65785112@news.neostrada.pl>
<k6tusp$elk$1@node1.news.atman.pl>
<5092979f$0$1208$65785112@news.neostrada.pl>
<k6u5vk$cf9$1@node2.news.atman.pl>
<5092a72c$0$1311$65785112@news.neostrada.pl>
<k6u98u$fjc$1@node2.news.atman.pl>
<5092aefd$0$1232$65785112@news.neostrada.pl>
<b...@g...com>
<5092c4d8$0$1222$65785112@news.neostrada.pl>
<k700o6$52t$1@news.task.gda.pl>
<50939638$0$26708$65785112@news.neostrada.pl>
<k71ct6$fjv$1@mx1.internetia.pl>
<5094e05f$0$1312$65785112@news.neostrada.pl>
<k73isd$sfj$1@mx1.internetia.pl>
<50958b34$0$26700$65785112@news.neostrada.pl>
<k783go$q3k$1@mx1.internetia.pl>
<5097a794$0$1299$65785112@news.neostrada.pl>
<k78amt$rth$1@mx1.internetia.pl>
<a...@n...plus.net>
<k79skj$dd7$1@node1.news.atman.pl> <k7aou5$c6s$1@news.task.gda.pl>
<k7ar3r$5oh$1@mx1.internetia.pl>
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <6...@g...com>
Subject: Re: B dny epsilon - this is not a bug, this is ?
From: Roman W <r...@g...com>
Injection-Date: Tue, 06 Nov 2012 12:32:31 +0000
Content-Type: text/plain; charset=ISO-8859-2
Content-Transfer-Encoding: quoted-printable
X-Received-Bytes: 2819
Xref: news-archive.icm.edu.pl pl.comp.programming:200527
[ ukryj nagłówki ]
W dniu wtorek, 6 listopada 2012 11:03:09 UTC użytkownik Michoo napisał:
> On 06.11.2012 11:31, AK wrote:
>
> > Poniewaz inline to tylko sugestia a nie oblig, wiec w praktyce moga (i
>
> > dzieja sie)
>
> > "cuda" np. w debug sa inne wyniki fp niz w release, niektore funkcje po
>
> > dodaniu
>
> > niewinnej linijki daja daja nieco inne wyniki itp itd.
>
>
>
> boost::Interval Arithmetic Library - wychodzi ~2x wolniej, ale pozwala
>
> całkiem nieźle oszacować, czy uzyskany wynik jest czymś więcej niż
>
> nagromadzeniem błędów obliczeniowych.

Da sie to "wrzucic" do starego kodu bez zmudnego przepisywania kazdej linijki?

RW