Time to complete project

Citrix · 2006-01-08, 09:54

Does anyone know how many years/hrs/min/ sec it will take to complete the project. THis will give us a better idea of how to proceed. How many machines we need etc.

In order to do so we need benchmarks of various n values. Do 100 iterations and calc average time per iterations. This need to be done, each time FFT length changes.

Anyone with expert knowledge on how to make these calculations?

Thanks,
Citrix

drakkar67 · 2006-01-08, 10:20

I think hunting primes in the near future is more important than the iteration times.
If we can hunt 1 or 2 primes until the n=2000000 barrier, it will be easy to recruit some more members (without losing some), and the computing power of the project will increase.
And, IMHO, density of the numbers to test is more important than iteration times. So, after we pass the n=2150000 barrier FFT length will increase and sieving will be more important. I hope the gap between 16000 and 85000 closes until then because sieving those ranges only eliminates candidates above 20M.
As a result, the first priority, IMHO, must be recruiting new members. And we must discuss how we can recruit them.
After we pass to 224 K FFT length (I think it will take 4-5 months) we will have a better idea (if we can eliminate some k's).

drakkar67

Citrix · 2006-01-08, 10:24

I am just interested in a rough estimate as what is the maximum time the project can take, if we do not find any prime till 50M. (Assuming we do not sieve any more)

Citrix

hhh · 2006-01-08, 15:55

Back of the envelope style.

We have 1.5 million tests until 50 M. Say we find some primes, sieve etc. and need to do only 500000 of them. One test takes one week on an average computer, given n=7M (some will take much longer, or a little less). So we have 500000 computer weeks, or 10000 computer years.
If we had 1000 computers, it would still take more than ten years to 50M.

.
The good news is: We might be done until then, too, while Seti will be still wasting electricity on

s.
Let's just continue crunching, and not too much worry about the future.
H.

NeoGen · 2006-01-08, 19:05

Quote:

Originally Posted by hhh

We might be done until then, too, while Seti will be still wasting electricity on

s.

I can't help to wonder if they would actually go public about it if they did happen to find some piece of alien communication...

I think not... so who knows if they haven't already picked up bits of it already?

Ok... I won't continue this line of thought because I know this would lead to an immense debate that is totally off-topic on a maths forum.

Just try to forget I said all that...

Brucifer · 2006-01-09, 03:45

Quote:

Originally Posted by NeoGen

Just try to forget I said all that...

heh.... okay....

2006-01-08, 09:54	#1
Citrix Jun 2003 3×5×107 Posts	Time to complete project Does anyone know how many years/hrs/min/ sec it will take to complete the project. THis will give us a better idea of how to proceed. How many machines we need etc. In order to do so we need benchmarks of various n values. Do 100 iterations and calc average time per iterations. This need to be done, each time FFT length changes. Anyone with expert knowledge on how to make these calculations? Thanks, Citrix

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Choosing a new correct time unit for the project	rogue	FermatSearch	6	2016-07-16 14:14
Expected Time To Complete A Quest Function	SaneMur	Information & Answers	33	2012-01-02 08:46
Total Project Time	paulunderwood	Five or Bust - The Dual Sierpinski Problem	2	2011-02-17 00:34
Time to complete information	JuanTutors	Software	3	2004-06-28 10:47
Shortest time to complete a 2^67 trial factor (no factor)	dsouza123	Software	12	2003-08-21 18:38

2006-01-08, 10:20	#2
drakkar67 Feb 2003 Istanbul 34₁₆ Posts	I think hunting primes in the near future is more important than the iteration times. If we can hunt 1 or 2 primes until the n=2000000 barrier, it will be easy to recruit some more members (without losing some), and the computing power of the project will increase. And, IMHO, density of the numbers to test is more important than iteration times. So, after we pass the n=2150000 barrier FFT length will increase and sieving will be more important. I hope the gap between 16000 and 85000 closes until then because sieving those ranges only eliminates candidates above 20M. As a result, the first priority, IMHO, must be recruiting new members. And we must discuss how we can recruit them. After we pass to 224 K FFT length (I think it will take 4-5 months) we will have a better idea (if we can eliminate some k's). drakkar67

2006-01-08, 10:24	#3
Citrix Jun 2003 3·5·107 Posts	I am just interested in a rough estimate as what is the maximum time the project can take, if we do not find any prime till 50M. (Assuming we do not sieve any more) Citrix

2006-01-08, 15:55	#4
hhh Jun 2005 373 Posts	Back of the envelope style. We have 1.5 million tests until 50 M. Say we find some primes, sieve etc. and need to do only 500000 of them. One test takes one week on an average computer, given n=7M (some will take much longer, or a little less). So we have 500000 computer weeks, or 10000 computer years. If we had 1000 computers, it would still take more than ten years to 50M. . The good news is: We might be done until then, too, while Seti will be still wasting electricity on s. Let's just continue crunching, and not too much worry about the future. H.