Okay. For what it's worth, I've never seen a segfault in sr1sieve or sr2sieve (the only variants I use) in several cpu-years-worth of PPC64 sieving. Do you know anything else it would do besides the segfault?

[QUOTE=BlisteringSheep;125608]Okay. For what it's worth, I've never seen a segfault in sr1sieve or sr2sieve (the only variants I use) in several cpu-years-worth of PPC64 sieving. Do you know anything else it would do besides the segfault?[/QUOTE]

The difference between addc and adde is that adde adds two registers plus the carry bit (stored in a different register) from a previous operation. addc ignores the carry bit. Both operations set the carry bit. In my case, the carry bit was 1 causing an invalid result from the mulmod which in turn lead to an infinite loop that eventually terminated when the program tried to access memory beyond the end of an array. This loop is used in setting up the discrete log. A miscalculation in this loop would cause it to miss factors fairly regularly. I re-ran the sieving I was doing from scratch and I haven't seen anything discrepancies with the factors it reports.

I'm not saying its impossible that no factors were missed or that invalid factors were output, but I am saying that it is less likely than the segfault.

Would it be possible to add the ability to convert to/from .dat format sieve files (i.e. Riesel Sieve, SoB, PSP) in srfile?

Thanks! :smile:

[QUOTE=rogue;125606]The specific fix is to replace the "adde" assembler instruction with "addc" in asm-ppc64.h and asm-ppc64.c. It might be in some other sources, but I haven't looked further yet.[/QUOTE]

I've made the adde --> addc changes in the latest sources. I agree that for sr1sieve, sr2sieve and sr5sieve a segfault is more likely than a missed factor. srsieve and gcwsieve probably wouldn't segfault however.

Does the -t switch work OK on OS X?

[QUOTE=geoff;125819]I've made the adde --> addc changes in the latest sources. I agree that for sr1sieve, sr2sieve and sr5sieve a segfault is more likely than a missed factor. srsieve and gcwsieve probably wouldn't segfault however.

Does the -t switch work OK on OS X?[/QUOTE]

I haven't tried it.

sr1sieve 1.3.4, sr5sieve 1.7.9
 

These versions fix a bug that can result in a Windows access violation when trying to print an invalid ETA date, which occurs if the ETA is too far in the future. Thanks to Chuck Lasher for reporting this one and for the help tracking down the cause.

Also fixed in the stable branch, sr5sieve 1.6.20, sr1sieve 1.2.8.

[QUOTE=Anonymous;125743]Would it be possible to add the ability to convert to/from .dat format sieve files (i.e. Riesel Sieve, SoB, PSP) in srfile?
[/QUOTE]

Yes this is possible, I'll add it to the list of things to do. (Another is to autodetect whether a .dat file contains +1 or -1 terms without needing the -s or -r switch).

[quote=geoff;127419]Yes this is possible, I'll add it to the list of things to do. (Another is to autodetect whether a .dat file contains +1 or -1 terms without needing the -s or -r switch).[/quote]
Thanks! :smile:

I hate to bug you for yet another feature so soon after the last one, but this one would come in really handy for various sieves I'm doing now and may do in the future.

You know srsieve's "warning" that it gives whenever a k has algebraic factors? Would it be possible to add a command line switch that will cause srsieve to automatically remove all candidates that are definitely not prime due to these algebraic factors (such as all even n for a k that is a power of 2, etc.)?

Thanks! :smile:

I have another request. Can sr(x)sieve be modified so that it can read the same input file formats that srfile can read?

Just the other day I ended up losing quite a bit of time on a sieve when a program unexpectedly crashed and took the PC with it. I didn't know where the sieve was up to, so I had to simply start from the last found factor (though I am reasonably confident it had progressed quite a bit further, I had no way of knowing how much further).

Therefore would it be possible to include the same checkpoint feature that sr2sieve has in sr1sieve? I would very much appreciate it and thank-you for your time and all the work you have put into this so far.