mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Octoproth Search (https://www.mersenneforum.org/forumdisplay.php?f=63)

 axn 2005-04-09 10:31

For n=82, there are 4 k's < 10^12

42290329515
481562533725
549711786105
624949113615

EDIT: No luck for n=81 for k < 10^12. This n was a "low weight" one compared to n=82

 ltd 2005-04-09 13:59

No luck for n=100,101,102.

@axn1 What OS are you using for your siever programm.
If it is windows is it possible that i can download it somewhere?

Lars

 robert44444uk 2005-04-09 14:34

no luck either

I did n=110 last night to k=10^11, no octoproths thier either

Regards

Robert Smith

 robert44444uk 2005-04-10 08:46

109

Checked 109 last night, one candidate that fell at the last hurdle, unlike the horse I chose for the Grand National, which cost me a tenner when it fell at the first!

Will do 108 tonight.

Regards

Robert Smith

 TTn 2005-04-11 02:51

interesting

Robert,

Nice find, I will have to check it out.
I have been touching up RMA a bit, and have'nt been online to catch up on what's going on.

TTn :grin:

 axn 2005-04-11 09:04

1 Attachment(s)
Results for n = 97 (after 65% completion to k<10^13)

1926973493115
2212009461375
2412877121565
5647136892825

@ltd - see the attached Pascal source code - Its not much. You'll have to modify the constants in the program and compile (you can use FreePascal compiler).

I plan to later clean it up and make it accept command line parameters :redface:

 ET_ 2005-04-11 09:43

[QUOTE=axn1]Results for n = 97 (after 65% completion to k<10^13)

1926973493115
2212009461375
2412877121565
5647136892825

@ltd - see the attached Pascal source code - Its not much. You'll have to modify the constants in the program and compile (you can use FreePascal compiler).

I plan to later clean it up and make it accept command line parameters :redface:[/QUOTE]

May I ask you which boost of performance gave the substitution of Pascal code with asm code?

Luigi

 axn 2005-04-11 10:11

[QUOTE=ET_]May I ask you which boost of performance gave the substitution of Pascal code with asm code?

Luigi[/QUOTE]

The two divisions in TestK - gave appr. 35% speedup. Not much but I'll take any speedup especially since I am running these for 2-3 days at a stretch. Actually, there is one more optimization there - the division of k by p is done by two back-to-back divisions; for most cases you only need one division. I plan to code it up and try it out. Let's see what kind of improvement it brings. For people needing non-asm version, you can use suitable qword operations. But in such cases, it might be worthwhile to use alternatives to division.

 axn 2005-04-11 11:38

[QUOTE=axn1]Results for n = 97 (after 65% completion to k<10^13)

1926973493115
2212009461375
2412877121565
5647136892825
[/QUOTE]

One more to the list for n = 97

6832047128535

 axn 2005-04-11 13:01

1 Attachment(s)
Uploading the latest version of the sieve along with the executable. The output needs to be redirected to some file. The resulting file can be further sieved using NewPGen.

 robert44444uk 2005-04-11 13:09

108

Ran 108 last night to 10^11, and no octos, sadly to say.

Axn1, will you be writing your code in a windows executable? I would certainly be interesting in devoting some raw computer power to take it further.

Regards

Robert Smith

All times are UTC. The time now is 12:56.