A few questions :) - Page 5

Vandy · 2002-09-02, 08:55

Ok, I know this is random but, who cares. Why aren't there any 2.52 Gigahertz comps on the benchmark page? I assume it just hasn't been updated in a while.

QuintLeo · 2002-10-04, 20:45

Benchmark page has had at least one update in the last week - my Athlon XP1800+ while it was at 1667.

Perhaps nobody is bothering to run the P-IV 2.53 at stock speed, of the few that have them?

lycorn · 2002-10-21, 19:37

This being a "questions" thread, just let me throw 2 more in:

1. Browsing the Status.txt file in the Primenet Server, I noticed that beejaybee has managed to crunch 61M iterations of 67108763 in ~3 months. It´s roughly 20M iterations/month, and with a large FFT! Anyone knows what machine is being used?

2. If you compare the 2 last Status pages in mersenne.org, you´ll notice that last week 517 exponents in the 69.1 - 79.3 M range have been factored. This is a huge amount of work

. Who is doing that and using what?...

smh · 2002-10-21, 20:52

Quote:

Originally Posted by lycorn

1. Browsing the Status.txt file in the Primenet Server, I noticed that beejaybee has managed to crunch 61M iterations of 67108763 in ~3 months. It´s roughly 20M iterations/month, and with a large FFT! Anyone knows what machine is being used?

20M iterations/month is more then 7 a second, which looks a bit to much, even for a P4 @ 3324 mhz (http://www.mersenne.org/overclk.htm)

IIRC beejaybee is Brain J. Beesly who is a Q&A tester of the program and a frequent poster on the mersenne mailing list, so maybe you should post your question there.

Quote:

Originally Posted by lycorn

2. If you compare the 2 last Status pages in mersenne.org, you´ll notice that last week 517 exponents in the 69.1 - 79.3 M range have been factored. This is a huge amount of work

. Who is doing that and using what?...

It's not as much as it seems. Numbers this high dind't have that much factoring work done (think something like 2^58 ). There are a few people who are factoring high numbers up to 2^60 or something like that. [/url]

binarydigits · 2002-10-21, 22:52

Quote:

Originally Posted by lycorn

1. Browsing the Status.txt file in the Primenet Server, I noticed that beejaybee has managed to crunch 61M iterations of 67108763 in ~3 months. It´s roughly 20M iterations/month, and with a large FFT! Anyone knows what machine is being used?

I am certain that Brian has been running that for much longer than 3 months. Until recently the server was ignoring status updates and results on QA work (or anything outside the server-assigned ranges). I did a lot of QA work myself (including DCing Brian's work) for which I got no credit from the server. Now for some reason the server seems to be acknowledging everything.

QuintLeo · 2002-10-23, 21:08

7 iterations a second is only around .140 - which is P-4 2Ghz ballpark, perhaps a little faster - though with the exponent size I suspect that the machine in question is more likely a 3-ish Ghz P-IV.

dragongoddess · 2002-10-24, 05:00

Question for you guys.
Does the machine that is running GIMPS need to have a 24/7 internet connection?

Deamiter · 2002-10-24, 05:08

Nope... actually it only needs to update once every 28 days or so ( actually about every 60 days would give you a good margin of error, but more than once every 28 days would be better) and of course every time your exponent gets finished. So basically, a dialup will work fine, it can be set to only access the internet when you've connected.

smh · 2002-10-24, 08:19

Quote:

Originally Posted by QuintLeo

7 iterations a second is only around .140 - which is P-4 2Ghz ballpark, perhaps a little faster - though with the exponent size I suspect that the machine in question is more likely a 3-ish Ghz P-IV.

According to the benchmark page a p4 2800M @ 3375M does 0,072 seconds per iteration for a FFT of 1792K

According to the status page a PII 400M takes 2,604/1,113 = 2,34 times longer for a 4096K FFT compared to a 1792K FFT

For a P4 @ 3375M this would approx. be 0,168 seconds per iteration as the best time. The average will be a bit higer. This means that this system can do approx. 15 million iterations a month which is about 25% less.

So a 2GHz machine will definately not show that much progress.

I think Binarydigits is right and this system is running the exponent longer then 3 months

Deamiter · 2002-10-24, 13:27

you gotta admit. they COULD be running on a cray.... think of the possibilities!!!

smh · 2002-10-24, 14:36

Quote:

Originally Posted by Deamiter

you gotta admit. they COULD be running on a cray.... think of the possibilities!!!

That would be an oldie

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Two questions:	Dubslow	GPU Computing	1	2011-08-05 18:22
Questions about the QS	Carmichael	Factoring	8	2007-04-10 11:30
5 questions	OmbooHankvald	Factoring	6	2005-08-28 19:31
Questions	OmbooHankvald	Prime Sierpinski Project	2	2005-08-01 20:18
LLR questions	OmbooHankvald	Math	6	2005-06-23 11:42

2002-09-02, 08:55	#45
Vandy Aug 2002 3 Posts	Ok, I know this is random but, who cares. Why aren't there any 2.52 Gigahertz comps on the benchmark page? I assume it just hasn't been updated in a while.

2002-10-04, 20:45	#46
QuintLeo Oct 2002 Lost in the hills of Iowa 2⁶×7 Posts	Benchmark page has had at least one update in the last week - my Athlon XP1800+ while it was at 1667. Perhaps nobody is bothering to run the P-IV 2.53 at stock speed, of the few that have them?

2002-10-21, 19:37	#47
lycorn Sep 2002 Oeiras, Portugal 2⁶·23 Posts	This being a "questions" thread, just let me throw 2 more in: 1. Browsing the Status.txt file in the Primenet Server, I noticed that beejaybee has managed to crunch 61M iterations of 67108763 in ~3 months. It´s roughly 20M iterations/month, and with a large FFT! Anyone knows what machine is being used? 2. If you compare the 2 last Status pages in mersenne.org, you´ll notice that last week 517 exponents in the 69.1 - 79.3 M range have been factored. This is a huge amount of work . Who is doing that and using what?...

2002-10-23, 21:08	#50
QuintLeo Oct 2002 Lost in the hills of Iowa 2⁶·7 Posts	7 iterations a second is only around .140 - which is P-4 2Ghz ballpark, perhaps a little faster - though with the exponent size I suspect that the machine in question is more likely a 3-ish Ghz P-IV.

2002-10-24, 05:00	#51
dragongoddess Oct 2002 9₁₀ Posts	Question for you guys. Does the machine that is running GIMPS need to have a 24/7 internet connection?

2002-10-24, 05:08	#52
Deamiter Sep 2002 1110101₂ Posts	Nope... actually it only needs to update once every 28 days or so ( actually about every 60 days would give you a good margin of error, but more than once every 28 days would be better) and of course every time your exponent gets finished. So basically, a dialup will work fine, it can be set to only access the internet when you've connected.

2002-10-24, 13:27	#54
Deamiter Sep 2002 3²·13 Posts	you gotta admit. they COULD be running on a cray.... think of the possibilities!!!