BOINC? - Page 7

Rytis · 2006-11-20, 20:18

I used newpgen on Celeron M 1.3 (my laptop), however, I am not planning to do any sieving (and I needed some for my reserved range to start testing ASAP), so I only sieved a few hours and only got quite a small number of candidates removed. This explains why the initial ranges had more candidates than the ones that gribozavr sent me.

gribozavr · 2006-11-20, 21:13

So you sieved 925-930 from scratch by yourself? If yes, then I understand everything. Sorry for the confusion.

pacionet · 2006-11-21, 13:49

Rytis it would be better if you update primegrid results also on twinprimesearch.org (manually) when you finished a range. (tis way we have an updated table).

Rytis · 2006-11-21, 16:36

I will; just that the first reserved range of 5M has not been fully finished yet (only 2M complete in that range, and other 3M vary between 56.6% and 69.17% complete). Or is it possible to report partial ranges (IE, smaller than initially reserved) too?

ValerieVonck · 2006-11-21, 17:28

Rytis,

Is it possible to have the following features:

- On your account page: number of primes found
- Link on the main page to the stats of LLR
- LLR Stats (drill down / range => who did the wu)

Thank you.

pacionet · 2006-11-21, 18:08

Quote:

Originally Posted by Rytis

I will; just that the first reserved range of 5M has not been fully finished yet (only 2M complete in that range, and other 3M vary between 56.6% and 69.17% complete). Or is it possible to report partial ranges (IE, smaller than initially reserved) too?

Uhm no .. you can report a range when you finished it.

Rytis · 2006-11-22, 17:20

Quote:

- On your account page: number of primes found

Done, if you have any primes found a new row is displayed below credit rows.

Quote:

- Link on the main page to the stats of LLR

I will do that once I release a public test version (ie, announce it on the front page). Probably Saturday.

Quote:

- LLR Stats (drill down / range => who did the wu)

It is already available in LLR range grid page (http://www.primegrid.com/orig/range_grid.php), however, only for the workunits that have primes found. I'm not sure if there is need to detail all workunits, as you may look your work done in your results page.

ValerieVonck · 2006-11-22, 17:55

Thank you!

KEP · 2006-11-22, 19:21

Will the release of a public test, mean that it will also be able to see how many LLR tests is queued and also to select to run just LLR, as one can do it at rieselsieve? (I didn't quite understood how though, maybe you can help). Also does the public test mean that we are going to see automatically workgeneration, and a larger range (at least 1 G) to be tested?

Rytis · 2006-11-26, 10:34

Hmm, for some reason I did not get email notifications of new posts here, and I thought that this thread has gone stale. Whoops :)

I started public test NOW (which still means that you have to select to run test apps, as I don't want to crash production systems by myself - I only allow people do that themselves ;)). I will add a link to range_grid.php today and will also add a new graph to charts page detailing how many unsent LLR results we have.

[edit]
Oh, and I forgot to mention that we finished the first reserved range of 5M with 2 primes found :)

Phased_Array · 2006-11-26, 13:14

2 obesrvations:

Since Rytis opened the LLR test to the public there has been an increase on the server load, I am getting the "unable to connect, server currently overloaded" error message more often then not. This problem will only increase as more people find out about the Twin Prime Search on PrimeGrid and start participating in the search.

Is it time to setup a donation drive for better hardware/more bandwidth?

Second, sieving. Is the sieving effort going to be able to stay ahead of the testing? The current setup allows one to run LLR testing only when work is available, dropping back to primegen when out of LLR tests. Not to take away from Rytis' project, but the die-hard prime searchers may not be comfortable with this setup.

Is a more concerted effort needed for sieving to stay ahead of the testing? Any thoughts on starting a second sieving effort on the next N value?

Anyway you look at it, this is a big break for the Twin Prime Search project, many thanks to Rytis for his help and efforts.

2006-11-26, 10:34	#76
Rytis Nov 2006 2³·11 Posts	Hmm, for some reason I did not get email notifications of new posts here, and I thought that this thread has gone stale. Whoops :) I started public test NOW (which still means that you have to select to run test apps, as I don't want to crash production systems by myself - I only allow people do that themselves ;)). I will add a link to range_grid.php today and will also add a new graph to charts page detailing how many unsent LLR results we have. [edit] Oh, and I forgot to mention that we finished the first reserved range of 5M with 2 primes found :) Last fiddled with by Rytis on 2006-11-26 at 10:37

2006-11-26, 13:14	#77
Phased_Array Aug 2006 7₈ Posts	2 obesrvations: Since Rytis opened the LLR test to the public there has been an increase on the server load, I am getting the "unable to connect, server currently overloaded" error message more often then not. This problem will only increase as more people find out about the Twin Prime Search on PrimeGrid and start participating in the search. Is it time to setup a donation drive for better hardware/more bandwidth? Second, sieving. Is the sieving effort going to be able to stay ahead of the testing? The current setup allows one to run LLR testing only when work is available, dropping back to primegen when out of LLR tests. Not to take away from Rytis' project, but the die-hard prime searchers may not be comfortable with this setup. Is a more concerted effort needed for sieving to stay ahead of the testing? Any thoughts on starting a second sieving effort on the next N value? Anyway you look at it, this is a big break for the Twin Prime Search project, many thanks to Rytis for his help and efforts. Last fiddled with by Phased_Array on 2006-11-26 at 13:31

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
BOINC	Unregistered	Information & Answers	6	2010-09-21 03:31
BOINC.BE	BATKrikke	Teams	2	2010-03-05 18:57
Boinc	Xentar	Sierpinski/Riesel Base 5	4	2009-04-25 10:26
BOINC?	masser	Sierpinski/Riesel Base 5	1	2009-02-09 01:10
BOINC	bebarce	Software	3	2005-12-15 18:35

2006-11-20, 20:18	#67
Rytis Nov 2006 58₁₆ Posts	I used newpgen on Celeron M 1.3 (my laptop), however, I am not planning to do any sieving (and I needed some for my reserved range to start testing ASAP), so I only sieved a few hours and only got quite a small number of candidates removed. This explains why the initial ranges had more candidates than the ones that gribozavr sent me.

2006-11-20, 21:13	#68
gribozavr Mar 2005 Internet; Ukraine, Kiev 11×37 Posts	So you sieved 925-930 from scratch by yourself? If yes, then I understand everything. Sorry for the confusion.

2006-11-21, 13:49	#69
pacionet Oct 2005 Italy 523₈ Posts	Rytis it would be better if you update primegrid results also on twinprimesearch.org (manually) when you finished a range. (tis way we have an updated table).

2006-11-21, 16:36	#70
Rytis Nov 2006 2³×11 Posts	I will; just that the first reserved range of 5M has not been fully finished yet (only 2M complete in that range, and other 3M vary between 56.6% and 69.17% complete). Or is it possible to report partial ranges (IE, smaller than initially reserved) too?

2006-11-21, 17:28	#71
ValerieVonck Mar 2004 Belgium 1517₈ Posts	Rytis, Is it possible to have the following features: - On your account page: number of primes found - Link on the main page to the stats of LLR - LLR Stats (drill down / range => who did the wu) Thank you.

2006-11-22, 17:55	#74
ValerieVonck Mar 2004 Belgium 7×11² Posts	Thank you!

2006-11-22, 19:21	#75
KEP Quasi Admin Thing May 2005 17×59 Posts	Will the release of a public test, mean that it will also be able to see how many LLR tests is queued and also to select to run just LLR, as one can do it at rieselsieve? (I didn't quite understood how though, maybe you can help). Also does the public test mean that we are going to see automatically workgeneration, and a larger range (at least 1 G) to be tested?