mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Prime Sierpinski Project

Reply
 
Thread Tools
Old 2009-02-10, 09:33   #1
opyrt
 
opyrt's Avatar
 
Apr 2008
Oslo, Norway

7×31 Posts
Default Factoring (again) ;-)

On my fairly OK computers, a PRP test takes approx. 36 hours with the "n" we're at now.

Is it time to reconsider factoring?
opyrt is offline   Reply With Quote
Old 2009-02-10, 11:38   #2
hhh
 
hhh's Avatar
 
Jun 2005

373 Posts
Default

Listen, nope.

Factoring is soon not even worth it anymore for SeventeenorBust. In half a year or a year, we could be there, i.e. sieved to 50P. Later, in three or four years, when SoB Firstpass reaches about 22M, it will become worth it again. For PSP, it's possible that we will never have to factor.
H.
hhh is offline   Reply With Quote
Old 2009-02-10, 13:29   #3
VJS
 
VJS's Avatar
 
Dec 2004

13×23 Posts
Default

hhh,

I would agree but do you have any hard evidence or numbers?

Opyrt, COuld I convince you to go back to secondpass. I'm running 3 cores on a set of numbers between 5M-6M so all we really need to do is push that secondpass level up to 5M. Once we get to 5M in that double pass we should get a huge jump to the firstpass level.

If we get one k to firstpass level I think it will help us alot in understanding or error rate. So far it looks very good but it's one of those things that you don't know until tested.
VJS is offline   Reply With Quote
Old 2009-02-10, 23:49   #4
opyrt
 
opyrt's Avatar
 
Apr 2008
Oslo, Norway

7·31 Posts
Default

hhh: OK, thanks. I did not mean to nag or anything, I asked just out of curiousity. If you felt my question was silly, it's most likely because my math skills don't quite grasp all this stuff. As far as understanding the different formulas for the primes and such, I understand. When it comes to if factoring is worth it or not, I don't have a clue. ;)
(well, now I do, as you answered me).

Quote:
Originally Posted by VJS View Post
Opyrt, COuld I convince you to go back to secondpass. I'm running 3 cores on a set of numbers between 5M-6M so all we really need to do is push that secondpass level up to 5M. Once we get to 5M in that double pass we should get a huge jump to the firstpass level.
I will return my quad to secondpass soon, but right now it's looking for wieferich primes. As you are a member of ARS Technica, I think I don't have to explain why... ;)
opyrt is offline   Reply With Quote
Old 2009-02-11, 00:07   #5
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

53·127 Posts
Default

Quote:
Originally Posted by opyrt View Post
I will return my quad to secondpass soon, but right now it's looking for wieferich primes. As you are a member of ARS Technica, I think I don't have to explain why... ;)
Please don't tell me that you are participating on this project. You are wasting your time. Geoff Reynolds and I supplied them with a core that would be at least 10x faster than what they are running, but they apparently dislike open source (our requirement for them to use our software was to open their source). I had already searched up to 3e15 without finding a Wieferich prime, although I did find a few "near" Wieferich primes. If you are working on that project it will be many years before it gets to 3e15.
rogue is offline   Reply With Quote
Old 2009-02-11, 00:54   #6
VJS
 
VJS's Avatar
 
Dec 2004

13×23 Posts
Default

rogue,

Please don't tell me this is true...

What your telling me is the same line Borkly feed us about a hacked client seti client back in the day.
The hacked client was identical (O.K. not identical b/c it was well written) that was 30-40% faster...
They said "don't use it... it hasn't been tested... the results are bougus... etc."

All along the only reason why they wouldn't release a faster client is they didn't have enough bandwidth to support the number of users returning results. Talk about client de-optimization.

At Orypt,

Yes I understand why your running the client and ARS wants 1st there, if Rouge is correct...

Run what you will and we appreciate your efforts.
Also a personal thanks for myself in running the DC que, you made a difference and were missed when you left.
VJS is offline   Reply With Quote
Old 2009-02-11, 08:40   #7
opyrt
 
opyrt's Avatar
 
Apr 2008
Oslo, Norway

21710 Posts
Default

Quote:
Originally Posted by rogue View Post
Please don't tell me that you are participating on this project. You are wasting your time. Geoff Reynolds and I supplied them with a core that would be at least 10x faster than what they are running, but they apparently dislike open source (our requirement for them to use our software was to open their source). I had already searched up to 3e15 without finding a Wieferich prime, although I did find a few "near" Wieferich primes. If you are working on that project it will be many years before it gets to 3e15.
First of all, thank you rogue. I am very grateful for people like you in the DC community that point out when something like this is happening. One of the main reasons that so many are working with that project is that they are in the DC-Vault.

Is there any chance that you can post about this on the forum of elmath.org? I am aware that you've already contacted them once, but this would be a kind of second chance. Also it would be out in the open for other members to see and comment on.

If they still don't want to swap their client for one that is 10x faster I see only one solution, and that is for AMD-Users, ARS Technica and Team Norway (the three top teams) to all stop running the project. And to join forces to get it removed from the DC-Vault. We are all kindly donating our CPU cycles and elmath.org should be humble enough not to waste them.

Quote:
Originally Posted by VJS View Post
At Orypt,
Yes, SJV? ;)
Quote:
Originally Posted by VJS View Post
Yes I understand why your running the client and ARS wants 1st there, if Rouge is correct...
Run what you will and we appreciate your efforts.
Also a personal thanks for myself in running the DC que, you made a difference and were missed when you left.
I promise you I will be back. If only llrnet would work on windows 2003 I would have had some other cores to add. (I've been in contact with the author and unfortunately, he does not have time to hunt down this bug). Manual reservations is not an options unfortunately.

Sorry for getting a bit off topic...
opyrt is offline   Reply With Quote
Old 2009-02-11, 11:08   #8
hhh
 
hhh's Avatar
 
Jun 2005

373 Posts
Default

Quote:
Originally Posted by opyrt View Post
hhh: OK, thanks. I did not mean to nag or anything, I asked just out of curiousity. If you felt my question was silly, it's most likely because my math skills don't quite grasp all this stuff. As far as understanding the different formulas for the primes and such, I understand. When it comes to if factoring is worth it or not, I don't have a clue. ;)
(well, now I do, as you answered me).
I really didn't mean to snob you or anything. Your question was posed a little funnily, so I thought I'd reply the same way. A smiley would have done a better effect, but I don't like smileys.

Cheers, H.
hhh is offline   Reply With Quote
Old 2009-02-12, 02:38   #9
VJS
 
VJS's Avatar
 
Dec 2004

13×23 Posts
Default

typoR,

Manual in DC is an option, but manual in DC is also quite the commitment. I've personally reserved a DC range with Lars form something like 5.5M-6M (with some missing I think).

Anyways that's about 100 days of work for me. But it will get done.

If you willing to make that kind of commitment Lars might run a special request for you. One of the great things about PSP is that you get what you ask for and the co-ordinators listen to the users, participate in the forums and take (and consider) suggestions.
VJS is offline   Reply With Quote
Old 2009-02-12, 08:13   #10
opyrt
 
opyrt's Avatar
 
Apr 2008
Oslo, Norway

D916 Posts
Default

Quote:
Originally Posted by VJS View Post
typoR,

If you willing to make that kind of commitment Lars might run a special request for you. One of the great things about PSP is that you get what you ask for and the co-ordinators listen to the users, participate in the forums and take (and consider) suggestions.
SVVJ: The issue is that I don't want to reserve such a range as there is quite a chance that it will end up like some of the other huge ranges that are reserved on this project; abandoned.

But what if we decide on a week where we focus on second pass? If you can try to talk Digital Concepts into this, I can try to get runesk to join?
(I meant diTigal Conpets and runSek )

We are usually the top producers on this project. Except for when Sloth is around... He can really make those WU's fly.

Oh... and we are waaaaay off topic again!
opyrt is offline   Reply With Quote
Old 2009-02-12, 16:29   #11
ltd
 
ltd's Avatar
 
Apr 2003

22×193 Posts
Default

Ok here are my answers to the on topic and off topic things in this post.

On topic. My tests also show that it will be a long time before P-1 will make sense again.

Off topic Second pass test:
@opyrt: If you like a reservation tell me. Reserving/unreserving such a test batch only needs less then 10 minutes so no problem at all. So if you have a set reserved and stop it only send in the results you have and say that you are stopping and thats it. For the private reservations on first pass I am in contact with the person who reserved them and he will send me some info which parts he wants to continue and which he wants to unreserve. But as usual there are real life restrictions for his time so we have to wait. But who am I to blame somebody for not having enough time. I am the maintainer of the project and I do not find the time to cleanup the "join a team" thread.

Another thing on the second pass server. I hope to get my i7 system tomorrow. I will let it do second pass tests for a while to find out how stable it is. Maybe after this I will switch to the low end of the second pass and try some overclocking.

And one more thing. It is quite possible that I will release some additional tests in the 4M-4.3M in the next four weeks. The incomming tests show two suspicious isers at the moment . If my local tests confirm this we will redo their work before we continue. But if this is the case I will send out a call for help.
ltd is offline   Reply With Quote
Reply

Thread Tools


All times are UTC. The time now is 17:34.


Sat Sep 24 17:34:26 UTC 2022 up 37 days, 15:03, 0 users, load averages: 0.61, 0.94, 1.03

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔