From: Frank Kotler on 28 Aug 2007 20:49 Betov wrote: .... > shhhhhtttt. I am still trying to understand why Herbert > would not make use of an Assembly Random, that could be > initialized on RDTSC... Where is the problem? Well... apparently Herbert wants it to work on machines that don't have rdtsc. I think I'd use "gettimeofday" (mov eax, 78). Returns a "structure"/"clump of bytes" (two dwords) (pointed to by ebx) indicating seconds and milliseconds since the "dawn of time". I'd take milliseconds, shift it up a bit, and xor it with seconds (to avoid an ever-increasing value), and use that as a seed. > Then, a good and simple RND we have found, right here, > was the delphi one: > ________________________________________________________ > > [Seed: Q$ 06227_8134] > > Proc DelphiMaxRandom: > Arguments @Max > Uses ecx, edx > > mov eax D(a)Max > imul edx D$Seed 0808_8405 > inc edx > mov D$Seed edx > mul edx > mov eax edx > EndP > ________________________________________________________ > > So... what? So... no Mersenne Twister??? :) I assume we initialize seed before using this (or we'll get the same sequence every time). Linux does provide /dev/rand, but otherwise the numbers don't know they're on Linux. Best, Frank
From: Betov on 29 Aug 2007 03:18 Frank Kotler <fbkotler(a)verizon.net> �crivait news:Ro3Bi.540$J65.106 @trndny08: > I assume we initialize seed before using this (or we'll get the same > sequence every time). No risk with Herbert LandScape... :)) Only the first view could be identical. > Linux does provide /dev/rand, but otherwise the > numbers don't know they're on Linux. As you said, any time function would do easily. But i still doubt there will be many users with a old 486 runing Linux, and, if there exist one, i wish him, that this would not be .... Ubuntu. :)) Betov. < http://rosasm.org >
From: Herbert Kleebauer on 29 Aug 2007 05:29 Betov wrote: > Frank Kotler <fbkotler(a)verizon.net> �crivait news:Ro3Bi.540$J65.106 > > I assume we initialize seed before using this (or we'll get the same > > sequence every time). > > No risk with Herbert LandScape... :)) > > Only the first view could be identical. No, with the same seed you always get the identical landscape. It doesn't matter how far you fly, all you see is only a 256x256 random field and the same one is used for the ground and the sky. > > Linux does provide /dev/rand, but otherwise the > > numbers don't know they're on Linux. > > As you said, any time function would do easily. But i still > doubt there will be many users with a old 486 runing Linux, > and, if there exist one, i wish him, that this would not be > ... Ubuntu. :)) I'm running Linux 1.18 on a 486 with 8 Mbyte RAM. And I have saved 8 identical PC's for replacement parts (needed only an ethernet card and two CPU fans for 13 years continous uptime).
From: Betov on 29 Aug 2007 05:52 Herbert Kleebauer <klee(a)unibwm.de> �crivait news:46D53C80.34ECFFFF(a)unibwm.de: >> No risk with Herbert LandScape... :)) >> >> Only the first view could be identical. > > No, with the same seed you always get the identical landscape. > It doesn't matter how far you fly, all you see is only a 256x256 > random field and the same one is used for the ground and the sky. Oh, yes, you are right... euuuuhhhh... no: I am wrong. :) > I'm running Linux 1.18 on a 486 with 8 Mbyte RAM. And I have > saved 8 identical PC's for replacement parts (needed only an > ethernet card and two CPU fans for 13 years continous uptime). Haven't you always been a bad example? :)) Betov. < http://rosasm.org >
From: CodeMonk on 29 Aug 2007 07:53
Herbert Kleebauer wrote: > > I'm running Linux 1.18 on a 486 with 8 Mbyte RAM. And I have > saved 8 identical PC's for replacement parts (needed only an > ethernet card and two CPU fans for 13 years continous uptime). 13 Years Continuous Uptime - WOW. I've got a spare razor if you need to trim those "tin whiskers" - it should be a mighty bearded beast by now. - Scott |