Maybe icc (I don't know that tool) defines __int64 so it does not like the line you show above. I would have to precede that line by #ifndef __int64.

I've just uploaded a new version. Please try it again.

Proxi problem ?
 

I've got the same files as I got for the first time.
Maybe it is a proxy problem and you should produce a different file at each new version ? like GENFERMv2.ZIP for fixing this ?

23698 oct 9 20:51 genferm.c
57344 oct 9 20:52 genferm.exe
33071 oct 11 10:46 GENFERM.ZIP

Tony

Tony,

I don't have the previous version, so please download the new one. The files should be:

11/10/2005 07:34 a.m. 24.795 genferm.c
11/10/2005 07:34 a.m. 57.344 genferm.exe
11/10/2005 08:27 a.m. 33.172 GENFERM.zip

The ZIP time and date depends on when you download the file.

Speedup in GENFERM.C
 

Since all prime factors of 2^(p^n)-1 are congruent to 1 or 7 mod 8 and the prime factors of 2^(p^n)+1 are congruent to 1 or 3 mod 8 (when p is an odd number), the program can be made faster by not performing the modular exponentiation when the candidate is congruent to 5 mod 8.

Another switch can be added to the program so it only computes factors of Generalized Fermat Numbers. In this case it can be made twice as fast as the current version because only candidates congruent to 1 or 7 mod 8 will be tested.

If p is 2 (for Fermat numbers), the prime factors of 2^(p^n)+1 are congruent to 1 or 5 mod 8.

I will try to post the new version during this week.

Are there any primes of the form?
(2^p^(e+1)-1)/(2^p^e-1) and (2^p^(e+1)+1)/(2^p^e+1)
With p=prime.
For sure Fermat primes fall into this category.
For p=3, there are a few small ones.
for 7^2, there is a twin prime ie. both terms above are prime
59^2 produces another prime

Are there any other primes of this form?


Citrix

[QUOTE=Citrix]Are there any primes of the form?
(2^p^(e+1)-1)/(2^p^e-1) and (2^p^(e+1)+1)/(2^p^e+1)
With p=prime.[/QUOTE] The conclusion of my discussion with Mr Cosgrave was that they only found one prime : [tex]\large \frac{2^{59^2}-1}{2^{59}-1}[/tex] of the first form. But he has lost the list of tested numbers and factors found.
Tony

did they miss 7^2?

I have tested all exponents under 50000 ie p^x<50000. May be we can extend this to 2 Million say, using the factor program. Are you interested?

citrix
The following are all the exponents.
[code]
9
25
27
49
81
121
125
169
243
289
343
361
529
625
729
841
961
1331
1369
1681
1849
2187
2197
2209
2401
2809
3125
3481
3721
4489
4913
5041
5329
6241
6561
6859
6889
7921
9409
10201
10609
11449
11881
12167
12769
14641
15625
16129
16807
17161
18769
19321
19683
22201
22801
24389
24649
26569
27889
28561
29791
29929
32041
32761
36481
37249
38809
39601
44521
49729
50653
51529
52441
54289
57121
58081
59049
63001
66049
68921
69169
72361
73441
76729
78125
78961
79507
80089
83521
85849
94249
96721
97969
100489
103823
109561
113569
117649
120409
121801
124609
128881
130321
134689
139129
143641
146689
148877
151321
157609
160801
161051
167281
175561
177147
177241
185761
187489
192721
196249
201601
205379
208849
212521
214369
218089
226981
229441
237169
241081
249001
253009
259081
271441
273529
279841
292681
299209
300763
310249
316969
323761
326041
332929
344569
351649
357911
358801
361201
368449
371293
375769
380689
383161
389017
390625
398161
410881
413449
418609
426409
434281
436921
452929
458329
466489
477481
491401
493039
502681
516961
528529
531441
537289
546121
552049
564001
571787
573049
579121
591361
597529
619369
635209
654481
657721
674041
677329
683929
687241
703921
704969
707281
727609
734449
737881
744769
769129
776161
779689
786769
822649
823543
829921
844561
863041
877969
885481
896809
908209
912673
923521
935089
942841
954529
966289
982081
994009
1018081
1026169
1030301
1038361
1042441
1062961
1067089
1079521
1092727
1100401
1104601
1125721
1129969
1142761
1181569
1190281
1194649
1203409
1216609
1225043
1229881
1247689
1261129
1274641
1295029
1324801
1329409
1352569
1371241
1394761
1408969
1419857
1423249
1442401
1442897
1471369
1481089
1495729
1510441
1515361
1530169
1560001
1585081
1594323
1630729
1635841
1646089
1661521
1666681
1682209
1692601
1697809
1708249
1739761
1745041
1760929
1771561
1852321
1868689
1874161
1885129
1907161
1953125
1957201
1985281

[/code]

[QUOTE=Citrix]did they miss 7^2 ? [/QUOTE] Yes, I've read again the part of his paper where he talks about his search. He may have not looked at p where Mersenne_p is prime.

"I decided to do try one small, Maple assisted check, and tested all ranks from 2 to 25; that is I tested the primes ... Of those twenty-four primes, fourteen lead to composite for the initial numbers in the corresponding generalised Fermat numbers, the composite Mersennes: ...
For each of those fourteen ranks I subjected the 1st level numbers (the ‘first cousins’ of those Mersenne numbers) to a base 3 Fermat test. Expecting all fourteen to be composite I was therefore expecting all fourteen to fail the Fermat base 3 test1. To my very, very great surprise I found that forthe 1031-digit number namely ..."

[QUOTE]I have tested all exponents under 50000 ie p^x<50000. May be we can extend this to 2 Million say, using the factor program. Are you interested?[/QUOTE] Why not ?! But my main problem is that I have very few free time and am interested by many things ... So I often do not finalize what Iv'e started to look at ...
If you build a table with p and e to test, yes I will record and run the factor program on the (p,e) I will reserve.
What about you alpertron ?
Tony

It would be better to change the program first to perform trial division only with the candidates that are congruent to 1 or 7 (mod 8) so it will tun twice as fast. I didn't have time to perform the change and upload the new version.

[QUOTE=T.Rex]

Why not ?! But my main problem is that I have very few free time and am interested by many things ... So I often do not finalize what Iv'e started to look at ...
If you build a table with p and e to test, yes I will record and run the factor program on the (p,e) I will reserve.
What about you alpertron ?
Tony[/QUOTE]

BTW you can use PFGW to test for these numbers. It is extremely fast. use -j9 or -j25 and so on to test each.