View Single Post
Old 2006-04-17, 21:17   #26
S00113's Avatar
Dec 2003

D816 Posts

Originally Posted by ewmayer
Sorry about the nomenclatural confusion (Peter Montgomery and I have last names beginning with the same letter, after all), but that's a different Mfactor than the one being discussed in this thread.
I should have guessed that. I have used your Mfactor before on AMD64, and noticed the output was very different. Great program btw. Once found a 70-bit factor for me.
As to the question about optimizations for 32-bit hardware, I have no plans for my mfactor code along these lines - just not worth the effort.
Completely understandable. I'm just doing this for fun. it is not high priority in any way.
Your same test example to 60 bits needs around a minute on decent 64-bit hardware running around 2GHz - even factoring in the clock speed difference that's nearly an order of magntiude faster in terms of per-cycle performance. My intention is not to discourage you from playing with this, but just to say that optimizing for 32-bit isn't where I'll be spending my own time.
My intentions with my wireless router is have fun and to fill gaps in factoring. Find more factors of exponents with known small factors, double check factoring from clients with bugs which may have caused them to miss a factor, etc. Trial factoring such composites is low priority work. Most, probably all, of the factors my slow wireless router find will already have been found by P-1, ECM or by trial factoring because the client wasn't buggy after all. I wouldn't use my precious 64-bit CPUs on this. Old 386s, 486s, PentiumIIIs, SGI Indys, DecStations and my cheap wireless router will do the job just fine while our modern machines use their resources on heavier crunching.
S00113 is offline   Reply With Quote