Exponent Size Gap

TimSorbet · 2007-03-22, 22:00

I've checked the 10 million digit number only box since I started GIMPS and here's the exponents I've received:
M33983237
M34354007 (~370 thousand gap)
M34657633 (~300 thousand gap)
M35041399 (~380 thousand gap)
M35431117 (just under 400 thousand gap)
M38125771 (~2.7 million gap)
The 35431117 one didn't take any longer than the others, but the gap difference is over 2 million. Is this because now even people that don't have the 10 million only box checked are getting 10 million digit exponents, so the amount that get done between the ones I do is a lot more?
I know this is probably the reason...but I'm just wondering why the gap is so huge.

tha · 2007-03-23, 09:23

Quote:

Originally Posted by Mini-Geek

I know this is probably the reason...but I'm just wondering why the gap is so huge.

The reason you stated is indeed a major part of the answer to the question. Also, as exponents get reassigned after being abandoned, you can sometimes get an exponent that is smaller than the previous one you obtained. So, there is a lot of 'noise' in the assignments.

davieddy · 2007-03-23, 09:37

About 6 months ago the "leading edge" of first time assignments
passed the 10,000,000 digit point and jumped to exponents~35,500,000
or so. So since then, it has moved faster in the 10,000,000+ digit range.

TimSorbet · 2007-03-23, 12:25

Quote:

Originally Posted by davieddy

About 6 months ago the "leading edge" of first time assignments
passed the 10,000,000 digit point and jumped to exponents~35,500,000
or so. So since then, it has moved faster in the 10,000,000+ digit range.

It was at ~35,500,000? That explains it. For almost the entire time between those two, anybody getting an assignment on first-time or 10-million checked out a number between those two.

davieddy · 2007-03-23, 15:33

How long do you take to check an exponent?

petrw1 · 2007-03-23, 16:47

It depends ... but I think the main factors are:
- Processor speed
- Allocated memory
- % of day you run Prime95

My 3.4Ghz takes about 28 days at the 35M range
My 2.4Ghz takes about 35 days at the 35M range
My now retired 400Mhz took 16 months at the 33M range
They both run nearly continuous.

TimSorbet · 2007-03-23, 17:30

Quote:

Originally Posted by petrw1

It depends ... but I think the main factors are:
- Processor speed
- Allocated memory
- % of day you run Prime95

My 3.4Ghz takes about 28 days at the 35M range
My 2.4Ghz takes about 35 days at the 35M range
My now retired 400Mhz took 16 months at the 33M range
They both run nearly continuous.

Mine's a 1533 Mhz Athlon XP with 140 MB allocated, and although I run it about 16 hours a day (the entire time I'm awake) for seven days a week, I use my computer and use it for some gaming.
I started the one I'm currently doing on August 14, 2006, didn't do TF, then finished the P-1 on August 24, 2006, and now it's going to finish the LL test today

, March 23, 2007. That's a day under 7 months for the LL. The last one took two days over 7 months for the LL, so I'm pretty consistently 7 months for LL in the 35M range. That makes it equivalent to about 9.5 hours each day.

So, davieddy, the short answer is seven months.

davieddy · 2007-03-23, 18:34

THX when I said how long does it take "you" I was talking
to Minigeek, not the world in general.
Personally, I am doublechecking 17,059,159 and am 93% through.
Long shot to find a prime tho

Taken ~ 6months on a PII.

tha · 2007-03-25, 07:29

All the exponents mentioned in the first post are exponents above the 10,000,000 digits range. But only part of the active GIMPS participants reserved exponents above the 10M digits range. (which is slightly above exponent 33,200,000) Now that we are out of first time LL assignments in the below 1)0M range all active participants reserve in the higher range most of the time. But even of more influence are the stress testing overclockers who reserve more than 50% of all assignments but never complete them. So the 'air bubble' that used to be in the below 10M range is now moving to the above 10M range. That explains most of the effect.

2007-03-22, 22:00	#1
TimSorbet Account Deleted "Tim Sorbera" Aug 2006 San Antonio, TX USA 10B7₁₆ Posts	Exponent Size Gap I've checked the 10 million digit number only box since I started GIMPS and here's the exponents I've received: M33983237 M34354007 (~370 thousand gap) M34657633 (~300 thousand gap) M35041399 (~380 thousand gap) M35431117 (just under 400 thousand gap) M38125771 (~2.7 million gap) The 35431117 one didn't take any longer than the others, but the gap difference is over 2 million. Is this because now even people that don't have the 10 million only box checked are getting 10 million digit exponents, so the amount that get done between the ones I do is a lot more? I know this is probably the reason...but I'm just wondering why the gap is so huge.

2007-03-23, 18:34	#8
davieddy "Lucan" Dec 2006 England 2·3·13·83 Posts	THX when I said how long does it take "you" I was talking to Minigeek, not the world in general. Personally, I am doublechecking 17,059,159 and am 93% through. Long shot to find a prime tho Taken ~ 6months on a PII. Last fiddled with by davieddy on 2007-03-23 at 18:37

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Max FFT Size?	aketilander	Software	44	2018-12-19 17:58
100M exponent FFT size?	NookieN	Information & Answers	3	2017-09-04 22:11
Pi(x) value for x at 10^16 size	edorajh	Computer Science & Computational Number Theory	6	2017-03-08 20:28
GHz days v exponent size	davieddy	Information & Answers	9	2011-07-16 20:50
FFT-Size	andi314	Lounge	14	2007-01-22 00:21

2007-03-23, 09:37	#3
davieddy "Lucan" Dec 2006 England 2·3·13·83 Posts	About 6 months ago the "leading edge" of first time assignments passed the 10,000,000 digit point and jumped to exponents~35,500,000 or so. So since then, it has moved faster in the 10,000,000+ digit range.

2007-03-23, 15:33	#5
davieddy "Lucan" Dec 2006 England 2×3×13×83 Posts	How long do you take to check an exponent?

2007-03-23, 16:47	#6
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 5²×211 Posts	It depends ... but I think the main factors are: - Processor speed - Allocated memory - % of day you run Prime95 My 3.4Ghz takes about 28 days at the 35M range My 2.4Ghz takes about 35 days at the 35M range My now retired 400Mhz took 16 months at the 33M range They both run nearly continuous.

2007-03-25, 07:29	#9
tha Dec 2002 2⁵·3³ Posts	All the exponents mentioned in the first post are exponents above the 10,000,000 digits range. But only part of the active GIMPS participants reserved exponents above the 10M digits range. (which is slightly above exponent 33,200,000) Now that we are out of first time LL assignments in the below 1)0M range all active participants reserve in the higher range most of the time. But even of more influence are the stress testing overclockers who reserve more than 50% of all assignments but never complete them. So the 'air bubble' that used to be in the below 10M range is now moving to the above 10M range. That explains most of the effect.