Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.lang.asm.x86    |    Ahh, the lost art of x86 assembly    |    4,675 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 3,374 of 4,675    |
|    wolfgang kern to Rod Pemberton    |
|    Re: Reciprocal MUL LUT    |
|    26 Apr 18 16:11:35    |
      From: nowhere@never.at              Rod Pemberton asked:              >> I try to shorten my current 512 bit 1/primes LUT              > What is the largest prime currently in your LUT? I.e., is this LUT       > very small, such as primes less than 100, or, is this LUT huge, such as       > having millions of primes?              work is still in progress ...       for now I test with only 16bit primes (largest: 0xfff1, count: 0x198e)              >> values for higher primes will really need 512 bits or more              > So, your LUT has 1/prime decimal fractions, converted to binary, which       > have periods that reach and exceed 512 bits?              Uint, Not decimal! dec 1/P periods are quite different to bin 1/P.              my first idea was to just use equal scaled 64 byte entries, but this soon       may become a heavy bloated story. So I check yet if different scaled periods       can save a lot on size w/o loosing too much speed.              > What is the smallest prime in your LUT which has a period of 512 bits?              haven't checked it yet, I have to write a period detector first.       __       wolfgang              --- SoupGate-Win32 v1.05        * Origin: you cannot sedate... all the things you hate (1:229/2)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca