I'm running with cuda:operations=8192 and filter:runlength=8192 now. > I do for sure notice that the computer is under heavy load... :) > Chains/s seems to go a bit above 152/s. But why is this rate rising > from the start to later stabilize? Are the other parameters for the > hardware autodetected, or should I try adjusting anything more?
the current rate is an average rate so it approaches the current rate in infinity (i.e. an hour). this will be fixed soon. you could try --device cuda:threads=512 if your GPU has enough registers, but last time i tried this, i got no improvement. > > My card is not overclocked, but can playing with these performance > parameters also give me wrong values? yes. my gtx260 runs at 691 mhz. it is simple to compute test values. use --work increment to use the same start values on every run, then with all other parameters the same, you should always get the same output. This way you should be able to verify the correctness of the computation. Write the data to a file or use --consume print:results=X where X is the number of chains you are generating, normally "print" only displays the first 16. to fuel to competition, my GTX260 with 216 cores does 162.54 chains/s at 691mhz, but since it is headless i do not have to care for responsiveness. > > As there is not that much messages on this list, I hope it's ok to be > a bit lazy and ask newbie questions without reading up or trying out > everything on my own... go ahead. the whole purpose of the project is education. (... and being in the newspapers :-) also take a look at the CommandLine page in the trac which should clarify some things. ________________________________________________________________ Neu: WEB.DE Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://produkte.web.de/go/02/ _______________________________________________ A51 mailing list [email protected] http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
