When do you think we will

[Joe O]

start handing out tests with n > 5000000?
I vote for November 3, 2003.

[MikeH]

I vote for November 1, 2003.

[Nuri]

I vote for October 31, 2003

[Mystwalker]

I bet on November 2nd.

btw. which time zone do we take? GMT?

[Nuri]

I used EST for October 31. It would be November 1st if GMT is used. :D

PS: This is to make it look like I made a precise estimation. :scratch: But it will probably fall way off anyways, we'll see. :blush:

[Keroberts1]

As passing 5 millioin is a meaning less mark how about how long till we find the next prime. Anyone have any guesses.

[Mystwalker]

Originally posted by Keroberts1
As passing 5 millioin is a meaning less mark how about how long till we find the next prime. Anyone have any guesses.

Considering the last couple of month, I vary from "when it rains cats and dogs" to "when hell freezes over". ;)
Frankly speaking, I hope soon, at least this year. A wholke year without a prime would be rather depressing, wouldn't it?

[eatmadustch]

The next prime is long overdue ...

[Joe O]

EST as defined on this stats page. (http://www.seventeenorbust.com/stats/rangeStatsEx.mhtml)

When will one or more of the Current test window Max n first show a value 5000000 or greater.

[Keroberts1]

I'm voting for 29th

[MikeH]

As long as the next prime is found while n<7M (possibly 8M), then it should be some time in 2004.

Still I agree, well overdue!

[Joe O]

Originally posted by MikeH
As long as the next prime is found while n<7M (possibly 8M), then it should be some time in 2004.

Still I agree, well overdue!

Since the difficulty, and the time required, increase O(n^2), the rate at which the max assigned n increases, has slown down (and will continue to slow down). I don't think that we will see n > 8M handed out in 2004, unless we find a prime!

[Keroberts1]

processor power has been increasing rapidly though either from increased membership or from the increases in processing power. Eithre way i am quite confident we'll finish 8 million long before the end of next year

[Joe O]

5M<n< 6M: 34265
6M<n< 7M: 33975
7M<n< 8M: 34075
Total 102315

In four months we have gone from 500 tests/day to 350 tests/day. This is in spite of the increasing membership and the increasing processor power. Remember the tests take longer with increasing n. It might be doable, it might not. We'll see.

[Mystwalker]

Are there some secret tests in the 500 tests/day? I know they had a great impact on the statistics...

[Keroberts1]

I would suspect that the secret tests had an impact i don't think i eve remember seeing 500 tests a day in the main active window.

[Joe O]

Originally posted by Keroberts1
I would suspect that the secret tests had an impact i don't think i eve remember seeing 500 tests a day in the main active window.
Where would you look for the tests/day "in the main active window"? The only number i am aware of would show a combined/total number. And yes it did spike when secret/supersecret were made available to everyone. The number was 500/550 a day before that and dropped to 450/day when all the hullaballoo was over. Now I didn't capture that information on a daily basis, but I have been keeping track of the n max test window since it was posted. You're welcome to the data and you can see for yourself that this rate of change has slacked off. This number would not be affected by the secret/supersecret tests. It was affected the time that someone grabbed up a whole lot of tests that subsequently expired. The rate zoomed thru the roof for a while and then slowed down to almost nothing. I stopped gathering statistics during this time. For those K that have been active all this year, the average for the year is around 500 n / day increase. Since all the K have been active, the average for the last 7 months is around 338 n /day.
Note that the project has been growing. As the graphs show, during the last 6 months we have gone from 150 cEMS/sec to 415 cEMS/sec. That is quite an increase, and shows no sign of letting up.


Now let's get back to the topic of this thread. Everyone is invited to post their best guess as to when the stats page will show that the n max active window has gone beyond 5000000 for one or more K.

[Nuri]

Originally posted by Mystwalker
Are there some secret tests in the 500 tests/day? I know they had a great impact on the statistics...

BTW, (almost) nobody is running secret anymore. I guess this is because it is the same limited number of (~30) people who are/were running secret, supersecret, seive, and P-1, and thanks to the change in sieve algorithm, more of the resources available to this group (= us) are now (~ for the last four weeks) shifted to sieving.

I'm sure most of you are aware of this. Just wanted to note it anyways...

[Nuri]

Now let's get back to the topic of this thread. Everyone is invited to post their best guess as to when the stats page will show that the n max active window has gone beyond 5000000 for one or more K.

PS: If anybody gets a 31337 (http://www.seventeenorbust.com/stats/users/user.mhtml?userID=6) test during this period, of course it will not count as we've gone beyond 5m. :looney:

Predictions so far:

October 29 = Keroberts1
October 30 = Eatmadustch
October 31 = Nuri
November 1 = MikeH
November 2 = Mystwalker
November 3 = Joe O

Anyone else?

[eatmadustch]

well, the 30th October isn't taken, I'll take that :D

[Moo_the_cow]

I vote for November 4 (I originally wanted to vote for November 2, but it was already taken :cry: and I think november 4th has a greater chance of suceeding than October 28th).

[MikeH]

:|ot|:

Since the difficulty, and the time required, increase O(n^2), the rate at which the max assigned n increases, has slown down (and will continue to slow down). I don't think that we will see n > 8M handed out in 2004, unless we find a prime! I'd agree, but only to a point. Although time required per test is related to n^2, this also means that moving from any given range size to the next requires a smaller increase in effort, thus

Candidates around n=3M require 125% more efort than for n=2M.
For 4M vs 3M it's 78%
For 5M vs 4M it's 56%
For 6M vs 5M it's 44%
For 7M vs 6M it's 36%
For 8M vs 7M it's 30%

Sieving and P-1 are still removing about 10 candidates per 1M n per day. With PRP moving at about 10,000 n per day (100 days per 1M n), that's 1000 less candidates required in each subsequent 1M n, so that 340 PRP tests per day comes down to 330, then 320. Yes, sieving will slow, but again not as much as you'd think.

So it's not as unrealistic as it may at first appear to keep moving along at about 10,000 n per day, and to reach n=8M by the end of 2004 (even without a prime).

Mike.

[Joe O]

Originally posted by MikeH
So it's not as unrealistic as it may at first appear to keep moving along at about 10,000 n per day, and to reach n=8M by the end of 2004 (even without a prime).
Mike.
Mike, I didn't mean to make it sound unrealistic, or even that I thought it was unrealistic to reach 8M in 2004. I should have posted my prediction. I think we shall reach 8M in January of 2005. But if your assumptions are correct, we could reach 8M in November of 2004, two months earlier. Finding a prime, would definitely speed things up though. <G> Joe.

[Joe O]

Originally posted by Nuri

Predictions so far:

October 29 = Keroberts1
October 30 = Eatmadustch
October 31 = Nuri
November 1 = MikeH
November 2 = Mystwalker
November 3 = Joe O

Anyone else?
November 4 = Moo the Cow

Well, It's lookinkg to be October 31/November 1. Now who is going to stay/get up to get the screen shot?

[Nuri]

I am wondering also, who will get the first (smallest) 5 million test (i.e. will he/she be someone using this forum or not, etc.).

[Joe O]

Originally posted by Nuri
I am wondering also, who will get the first (smallest) 5 million test (i.e. will he/she be someone using this forum or not, etc.).
Nuri,

That deserves a thread of it's own. Ask everyone to post their smallest 5 million test. At least we'll find out who from the forum has the smallest. Unless MikeH has another way.

[Nuri]

I'm planning to do it (waiting for my first 5 million test). If anyone else posts such a thread before I do, that's also fine.

Thanks to P-1, it is very easy to find out which k/n pair is actually the smallest above 5 million.

I've not checked with the latest results.txt, but as of the results.txt two weeks ago, the smallest three candidates are:

4847/5000007
24737/5000023
55459/5000026

To determine which user got it, however, would be hard I guess (unless it's someone from the forum who wants to announce it, or Louie announces it). Anyways, we'll see.

[Joe O]

Does anyone wnat to guess when on the 1st we will start handing out n > 5000000?

I'm guessing between midnight and 4 am.

[Nuri]

I'm sticking with my guess of October 31 if EST is used and November 1st if GMT is used.

If I'm not mistaken, there should be five hour difference. So, it's October 31st, from 19:00 to midnigt EST.

[Joe O]

Originally posted by Nuri
I'm sticking with my guess of October 31 if EST is used and November 1st if GMT is used.

If I'm not mistaken, there should be five hour difference. So, it's October 31st, from 19:00 to midnigt EST.

Well, you are still very much in the running. It's you and MikeH at the moment. Either the last 4 hours of the 31st or the first 4 hours of the 1st, using EST.
I would have much preferred to use GMT or UTC, but Louie's page is in EST so I've stuck with that.

[Joe O]

Nuri,
"What a difference a day makes"
Now it is looking very solid to be the 31st. Somewhere between 6am and 6pm.

[Keroberts1]

10:24 GMT oct 31st

[pcdeatherage]

55459-5001358 10.31.03 07:57 (texas time)

(oops, sorry, didn't see Keroberts1 post above)

[MereMortal]

I just got k=27653, n=5001405 (5:44 PST)