Author: Gerd Isenberg
Date: 09:42:27 12/25/05
Go up one level in this thread
On December 25, 2005 at 11:54:16, Vincent Diepeveen wrote: >On December 25, 2005 at 11:39:12, Gerd Isenberg wrote: > >>On December 25, 2005 at 11:08:45, Vincent Diepeveen wrote: >> >>>On December 25, 2005 at 10:27:34, Gerd Isenberg wrote: >>> >>>Hello Gerd, >>> >>>Please note that the lastname 'De Groot' is very common in chess. Like the name >>>'ye' in china or the name 'muh. >>> >>>But please realize this Frank de Groot is completely disturbed person. >> >>Aha - i didn't got the impression while reading. >>Found the "Secret Opcode" article quite amusing. >> >>Gerd > >Please confuse this guy not with other De Groot's. There is many De Groot's in >computerchess actually. They are completely different persons. Not even family >from this guy. There is in fact other Frank de Groot persons who are in >computerchess. Don't confuse them with this go guy. I only quoted the Author's name!? It is likely that there are other Frank de Groots i am not aware of ;-) > >Vincent > >p.s. for factorization or encrypting/decrypting whatever, >counting number of bits is not so interesting >more interesting is fast multiplication of numbers. that gets however >done also in *hardware* not in software. they use special chips for that. >such chips are a lot faster than you can do in software >for those applications. A single hardware chip is faster than teraflop >supercomputers. > >so suppose i design a new factorization search using artificial intelligence, >then i'll have to convince some dudes there to get a funding and then they'll >put it in hardware. software is too slow for that. > >it involves multiplying numbers a lot and sometimes modulo calculations. >That multiplying can be done for example with FFT. >that requires only multiplications, in hardware they use a different algorithm >than fft - i am not an expert in those respects, but bitcounting is not so >interesting. > >requirements for chips are therefore interesting and i'll always shout for chips >with faster multiplication capabilities, but really for a pc processor a fast >popcount is just irrelevant. Setwise interpretation of computerwords is quite common - not only for bitboards. So is getting the cardinality of a (sub-)set. Gerd > >for those applications hardware is so much faster than software, that there is >not even a discussion about software implementation of an algorithm. It's only >about how to get it in *hardware*. > >> >>> >>>Vincent >>> >>>>Nice blog on "Evolution of a Go program" from Frank de Groot: >>>>http://www.moyogo.com/blog/blog.htm >>>> >>>>I found following article on popcount instruction particular interesting: >>>>http://www.moyogo.com/blog/2005/09/secret-opcodes.html >>>> >>>>The answer from Christian Ludloff on my question in sandpile forum some time ago >>>>is still a mystery to me: >>>> >>>>"PS: Don't call it POPCOUNT. The proper mnemonic would have BT >>>>at the beginning, though BTC is already taken..." >>>> >>>>Gerd
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.