| Author |
Message |
|---|
|
|
|
As of 4:30 pm PDT July 4, I am the person with the most credit but *zero* primes on the PPSElow port. No worries... Once again, into the breach...
--Gary |
|
|
|
|
|
Speak of the devil... not so much as an hour after I posted this, I got my first hit on the PPSElow port... 2223*2^224808+1. Well I had that ignominious distinction for a while :-) |
|
|
 Neo Volunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
                   
|
|
Gary,
Don't feel bad, I think I just had an 8K w/u stretch with no prime on the PPSeLow port, yet, the average is about 2.3K tests per prime. So...
Anywh00t, congrats on your prime. Saw that you made today's prime finder list. :)
Neo |
|
|
|
|
|
Yeah they come and go. I had one stretch last week of like 15k without a prime... And I know rick had a slow day a couple days ago too.
Although on the other hand, I also had one day where I had 3 primes in 4 minutes. :-D It comes and goes. Luck of the draw I suppose.
Good luck! Hopefully the next one comes sooner. |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
Gary,
How are you pulling down your w/u's? (in what quantities?)
Most of the time I have each core pull down a different number of w/u's at a time. One core will pull down 7 while another core will pull down 17, and even a core pulling 3 down at a time. Rick taught me this strategy and it seems to work (the not putting all your eggs in one basket trick).
Neo |
|
|
|
|
Gary,
How are you pulling down your w/u's? (in what quantities?)
Most of the time I have each core pull down a different number of w/u's at a time. One core will pull down 7 while another core will pull down 17, and even a core pulling 3 down at a time. Rick taught me this strategy and it seems to work (the not putting all your eggs in one basket trick).
Neo
You know, when I heard Rick talk about adjusting the number of wu's I didn't listen. Later after no primes I switched strategies and hit the primes. I was pulling down 20-15-10-5.
Anyway, congrats Gary!! Stick with it, and they will come. Though I do have a theory that primes avoid i7 2600K processors....
For those of you who understand such things, has anyone read any of Terence Tao's papers on the random nature of primes? |
|
|
|
|
|
i had 10 cores pulling down in 20 at a time, and 6 cores doing 5 at a time.
It made no difference for me. I was fairly evenly getting primes across all three computers. If you do enough work units regardless, you are bound to get a lot of primes. |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
i had 10 cores pulling down in 20 at a time, and 6 cores doing 5 at a time.
It made no difference for me. I was fairly evenly getting primes across all three computers. If you do enough work units regardless, you are bound to get a lot of primes.
Mark,
I am assuming you don't know this about prpnet: If you pull w/u's down in composite numbers (i.e., 4, 10, 20), the server will send you double check work only. You will notice that all your primes on the low port came from your prpclients that were pulling down 5 w/u's at a time. 5 is prime.
Sorry, I thought you knew.
Neo |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
Ok,
I jest.
Just thought I would raise your blood pressure 20 points temporarily :)
Neo |
|
|
|
|
i had 10 cores pulling down in 20 at a time, and 6 cores doing 5 at a time.
It made no difference for me. I was fairly evenly getting primes across all three computers. If you do enough work units regardless, you are bound to get a lot of primes.
Mark,
I am assuming you don't know this about prpnet: If you pull w/u's down in composite numbers (i.e., 4, 10, 20), the server will send you double check work only. You will notice that all your primes on the low port came from your prpclients that were pulling down 5 w/u's at a time. 5 is prime.
Sorry, I thought you knew.
Neo
Good thing I've been around prpnet and you longer than Mark cause I was 1/2 way through this before I realized you were trying to make Mark do a lot of busy work checking on his primes :)
That was very good there Neo!
____________
@AggieThePew
 |
|
|
|
|
|
haha very nice! the sad part I believed you, but was more trying to figure out how i DID have primes from my clients that pull down 20. :-)
You win. You successfully confused the computer analyst. |
|
|
|
|
Ok,
I jest.
Just thought I would raise your blood pressure 20 points temporarily :)
Neo
Should that be 19 points, 19 is prime right?
Neo, you are the man, you got me. I had to read it twice as I nearly believed it the first time
____________
Welcome to Holland
 |
|
|
|
|
Ok,
I jest.
Just thought I would raise your blood pressure 20 points temporarily :)
Neo
Should that be 19 points, 19 is prime right?
Neo, you are the man, you got me. I had to read it twice as I nearly believed it the first time
I've got a nice bridge in Brooklyn for sale if any one is interested... |
|
|
|
|
I've got a nice bridge in Brooklyn for sale if any one is interested...
Really? How much?
P.S. Neo had me too for a bit..... |
|
|
|
|
|
I think we are all too trusting here :-) Lesson learned, haha, don't trust Neo.
just kidding. That was really good tho! |
|
|
|
|
|
Well there's a lot of Sturm und Drang on the main boards about the PPS Sieve credit, but I'm just glad to have that nasty '0' in my PPSELow prime count changed to a '1'.
As for my download "strategy", for short tasks like the Low port or SGS, I tend to grab a fair number of WUs at a time, figuring that the network traffic will be more efficient that way (less time spent in the protocol overhead). For longer units like Mega or the GFN GPU tasks, the WU download time is negligible compared to the run time, so I just get them one by one. It would surprise me if the d/l size has an effect on the chances of any particular task turning out prime. The "prime number theorem" (see wikipedia) states (roughly) that the chance of a randomly chosen number being prime is inversely proportional to its number of digits (base 'e'). So grabbing 2 blocks of 5 WUs vs. 1 block of 10 should give me equal chances, assuming the numbers being tested are the same, and barring some sort of weird bug in the s/w distributing the tests.
I've actually moved off PRPnet for a few days, doing some boinc cw & trp sieving for a while.
--Gary |
|
|
|
|
|
First, congrats on your prime Gary. Second, remember I said when I first stated my strategy that it was all seat of the pants :) What first started me looking at different caches for different pc's was the run time. I noticed that some wu's took much longer to run depending on size and I didnt want my slow pc's hung up on the longer ones while other people were getting the short runs so I sized my caches to my pc's. I know, makes no sense but that's why I'm a member of Aggie, we do things cause we want to :)
____________
@AggieThePew
|
|
|
|
|
|
Any strategy that gets the occasional hit and doesn't set the house on fire is a reasonable one in my book. :-) Keeping all those machines efficiently fed must be a fuss; I've only got two (sometimes three) to worry about.
Everyone here is doing great work!
--Gary |
|
|
|
|
Well there's a lot of Sturm und Drang on the main boards about the PPS Sieve credit...
--Gary
Surely not suggesting I've been throwing fat in that fire? :)
|
|
|
