... I could probably do that, but it will generate a lot of trace. As to what the bug is dependent on, I don't know. All I know is that some nodes have a...
2351
Phil Carmody
thefatphil
Feb 9, 2013 10:58 am
... Ugh, I'd forgotten how much I dislike C++ and Systems Hungarian notation. THe only thing that jumped out at me was this: unsigned til =...
2352
Phil Carmody
thefatphil
Feb 9, 2013 11:15 am
... That means it's the twin-erat scenario, the code path I quoted, doesn't it? I'm sure David is looking forward to testing 3.7.3 later today... ;-) Phil...
2353
Mark Rodenkirch
mgrogue
Feb 9, 2013 2:50 pm
... Thanks Phil. I would not have been able to figure that out. ./pfgw64 -f'500{7921,+-1}' -d...
2354
djbroadhurst
Feb 12, 2013 9:27 pm
Thanks to Mark for version 3.7.3 for win and mac. This tester awaits 'nix, avidly. David...
2355
mgrogue
Feb 12, 2013 10:10 pm
... I sent an e-mail to Steven Harvey on Sunday (who builds for Linux as I can't), but he hasn't gotten back to me. --Mark...
2356
StevenH
harvey563
Feb 13, 2013 6:07 pm
It's up now. Sorry for the delay. Steven Harvey...
2357
djbroadhurst
Feb 14, 2013 2:38 am
... Sadly, Steven's linux build is still broken. Here is the correct behaviour, with version 3.6.7: $ pfgw367 -f'{7921,+-1}' -N -d -e100000000000...
2358
Mark Rodenkirch
mgrogue
Feb 14, 2013 3:34 am
... I just did a retest. The 32-bit Mac pfgw works correctly. The 64-bit Mac version gets an FPE, but it works correctly when compiling without...
2359
djbroadhurst
Feb 14, 2013 11:20 am
... No, it too misses the factors: $ pfgw32s -f'{7921,+-1}' -N -d -e100000000000...
2360
Phil Carmody
thefatphil
Feb 14, 2013 4:56 pm
... Does +0.0039s mean that factoring stopped almost immediately? ... +++ pform/pfoo/treefactorize.cpp (revision 594) @@ -373,8 +373,8 @@ unsigned til =...
2361
Phil Carmody
thefatphil
Feb 14, 2013 8:34 pm
OK, I've logged onto an x86_64 machine: openpfgw-code$ make make -C pform/pflib make[1]: Entering directory...
2362
mgrogue
Feb 14, 2013 8:39 pm
... I had removed configure a while ago since it didn't do much. I have yet to fix the makefile and put the necessary files where needed in order for anyone...
2363
StevenH
harvey563
Feb 15, 2013 4:43 pm
I get the same results on each: steven@steven-desktop5 ~/op37 $ ./pfgw64sg -f'{7921,+-1}' -N -d -e100000000000...
2364
Mark Rodenkirch
mgrogue
Feb 18, 2013 12:41 pm
... Another missed factor was reported on mersenneforum. Phil is helping me track down the problem. --Mark...
2365
djbroadhurst
Mar 11, 2013 10:48 pm
... It seems, from my notification and this later one, that http://sourceforge.net/projects/openpfgw/files/ has now been dysfunctional for several weeks. No...
2366
Mark Rodenkirch
mgrogue
Mar 11, 2013 11:33 pm
... I don't know why that is dysfunctional. It works for me. I could revert back and remove the tree factoring code. It doesn't seem to me that I'll be able...
2367
djbroadhurst
Mar 11, 2013 11:59 pm
... I would, also. Thanks for the factoring speed-up that was sometimes available, on some platforms. But in the interest of wider reliability, I'd recommend a...
2368
Phil Carmody
thefatphil
Mar 12, 2013 9:01 am
No idea what's happening with attributions, yahoo doesn't like your mailer... ... I'd hardly say I was the expert. I don't write C++, didn't write PFGW, and...
2369
Mark Rodenkirch
mgrogue
Mar 12, 2013 11:41 am
... http://www.mersenneforum.org/gimps/ Who did contribute the code? --Mark...
2370
djbroadhurst
Mar 15, 2013 8:20 pm
... Jim wrote the tree-sieve code: http://tech.groups.yahoo.com/group/openpfgw/messages/1580 David...
2371
Mark Rodenkirch
mgrogue
Mar 22, 2013 10:56 pm
3.7.4 is ready to go. I removed the tree factoring code (too many issues) and fixed a couple of other bugs. The Mac and Windows builds are posted. You will...
2372
djbroadhurst
Mar 24, 2013 8:24 pm
... Sadly, the factoring is still broken. Here is the correct behaviour with v3.6.7: $ pfgw367 -f'{7921,+-1}' -N -d -e100000000000...
2373
Mark Rodenkirch
mgrogue
Mar 24, 2013 10:17 pm
I'm sorry. I screwed up. I thought I had reverted to a non-treefactoring version of code. I was wrong. In the process I found another problem that affects...
2374
paulunderwooduk
Mar 25, 2013 8:34 pm
... Version 3.7.5 is at http://sourceforge.net/projects/openpfgw/ ./pfgw64 -f'{7921,+-1}' -N -d -e100000000000...
2375
djbroadhurst
Mar 25, 2013 10:19 pm
... Thanks, Mark and Steven, for the now successful reversion to the factoring code of v.3.6.x, in v3.7.5, whose statically linked 64-bit linux executable has...
2376
djbroadhurst
May 4, 2013 11:10 am
... More rigorous testing revealed a factoring bug. This is correct: $ pfgw -f'{18773,-1}' -d -e3000000000 -q'(2*18773*57203-1)*(2^63+29)' PFGW Version...
2377
Mark Rodenkirch
mgrogue
May 4, 2013 1:00 pm
... I've taken a brief look at the code, but not enough to determine the cause of the problem. I'm very busy right now with work so my focus has been on that...
2378
djbroadhurst
May 5, 2013 8:25 pm
... For future reference, I believe that it is significant that the missing factor 2*18773*57203 - 1 = 2147743837 = 2^31 + 260189 is only slightly above 31...
2379
djbroadhurst
May 6, 2013 8:46 am
... Here is another such case, where the factor is missed using the +-1 option $ pfgw -f'{20269,+-1}' -d -e3000000000 -q'(2*20269*52981-1)*(2^63+29)' PFGW...
