On 14 Feb 2002, at 19:44, Michael Vang wrote: > > I doubt George would be interested in working in a little simple zip > > routine when saving/reading save files? It might slow it down too > much > > for some folks (although honestly, it takes a fraction of a second to > > zip using the lowest compression level), but maybe a nice option for > > those looking to save a bit of space when testing large exponents. > > Especially if you save interim files or have 2 saved files... the > space > > savings would add up quickly. > > My vote goes towards leaving Prime95 just the way it is... I prefer to > have a program do one thing well rather than have it try to do > everything under the sun...
My vote too. It's so easy to "manually" compress files with a script if _you_ really think it's worthwhile. FWIW how effective zip is depends on where the exponent is in its FFT run length size range. You get significantly better compression for exponents near the low limit than for exponents near the high limit. The reason should be fairly obvious. FWIW bzip2 does a significantly better job of compressing save files than zip, but at the expense of using several times as many CPU cycles. I'm doing a QA run on exponent 67108763 keeping interim files at million iteration intervals, so, by the time I finish, I will have 67 save files. The raw size is 14MB per file, but bzip2 reduces them to around 9MB. Whether we should be unduly worried about a job which takes ~ 1 year to run producing ~ 1GB of data, in these days when it's hard to buy new HDDs smaller than 20GB, is a matter of opinion. FWIW Glucas save files are stored in packed integer format; the processing overhead seems to be reasonable & the raw files produced by Glucas are almost exactly the same size as a bzip2'd Prime95/mprime save file for a similar exponent. However even bzip2 will not compress Glucas save files further. The reason should be obvious - the information has to be stored somewhere, and lossy compression techniques are obviously of no use in this particular application! > > I personally would like to see Prime95 go the other direction to a > simpler CLI interface, like Mlucas, but I'm sure I'm in the minority... My major criticism would be that, at the moment, there are some fairly important tweaks you can do by editing the ini files, but not through the menu system. It can be confusing as to what needs to happen to get a setting changed on a running system - does it happen automagically, do you have to stop & continue or do you have to exit & restart the program to take effect? mprime is much closer to a traditional CLI interface. linux rules:) Regards Brian Beesley _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
