dieharder dla dużych wartości m, np. m=1000 - Grupy dyskusyjne w eGospodarka.pl

eGospodarka.pl › Grupy › pl.comp.programming › dieharder dla dużych wartości m, np. m=1000 › dieharder dla dużych wartości m, np. m=1000

X-Received: by 10.31.180.138 with SMTP id d132mr768813vkf.9.1513173832508; Wed, 13
Dec 2017 06:03:52 -0800 (PST)
X-Received: by 10.31.180.138 with SMTP id d132mr768813vkf.9.1513173832508; Wed, 13
Dec 2017 06:03:52 -0800 (PST)
Path: news-archive.icm.edu.pl!news.icm.edu.pl!newsfeed2.atman.pl!newsfeed.atman.pl!go
blin2!goblin.stu.neva.ru!peer02.am4!peer.am4.highwinds-media.com!peer03.iad!fee
d-me.highwinds-media.com!news.highwinds-media.com!m31no311745qtf.0!news-out.goo
gle.com!v55ni218qtc.0!nntp.google.com!m31no311742qtf.0!postnews.google.com!gleg
roupsg2000goo.googlegroups.com!not-for-mail
Newsgroups: pl.comp.programming
Date: Wed, 13 Dec 2017 06:03:52 -0800 (PST)
Complaints-To: g...@g...com
Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=178.37.224.118;
posting-account=xjvq9QoAAAATMPC2X3btlHd_LkaJo_rj
NNTP-Posting-Host: 178.37.224.118
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <c...@g...com>
Subject: dieharder dla dużych wartości m, np. m=1000
From: "M.M." <m...@g...com>
Injection-Date: Wed, 13 Dec 2017 14:03:52 +0000
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: base64
X-Received-Body-CRC: 2408864377
X-Received-Bytes: 3813
Xref: news-archive.icm.edu.pl pl.comp.programming:211869
[ ukryj nagłówki ]
Dieharder chyba pada na dużych wartościach m.

Testu dnia urodzin nie przechodzi generator ranlux48, ani mój generator
fibonacciego z poprzedniego wątku:

FIBONACCI
time ./frnd | dieharder -d 0 -m 1000 -g 200
#===================================================
==========================#
# dieharder version 3.31.1 Copyright 2003 Robert G. Brown #
#===================================================
==========================#
rng_name |rands/second| Seed |
stdin_input_raw| 4.97e+06 |1997299816|
#===================================================
==========================#
test_name |ntup| tsamples |psamples| p-value |Assessment
#===================================================
==========================#
diehard_birthdays| 0| 100| 100000|0.00000014| FAILED

real 34m8.515s
user 35m39.480s
sys 1m20.016s

RANLUX48
time ./frnd | dieharder -d 0 -m 1000 -g 200
#===================================================
==========================#
# dieharder version 3.31.1 Copyright 2003 Robert G. Brown #
#===================================================
==========================#
rng_name |rands/second| Seed |
stdin_input_raw| 3.78e+05 |1785358803|
#===================================================
==========================#
test_name |ntup| tsamples |psamples| p-value |Assessment
#===================================================
==========================#
diehard_birthdays| 0| 100| 100000|0.00000013| FAILED

real 93m39.541s
user 124m16.952s
sys 0m41.020s

Nie wiem co jest grane, ale chyba brakuje precyzji dieharderowi, rozkłady
powinny być zaimplementowane chociaż na precyzji 1024 bitów.

Pozdrawiam

P.S.
Uważajcie gdy patrzycie na czasy, bo to nieprawda że ranlux działa tylko 3 razy
wolniej niż fibonacci.