BOINC? - Page 16

jasong · 2006-12-22, 01:41

Quote:

Originally Posted by Rytis

This might just mean that the LLR final beta is just around the corner (probably during the week between Christmas and New Year).

Skligmund · 2006-12-22, 03:07

Quote:

Originally Posted by jasong

Which brings me to my next point: Don't Do Crack.

jasong · 2006-12-22, 20:53

Quote:

Originally Posted by Skligmund

Don't Do Crack.

/me suddenly remembers to pull up his pants.

jmblazek · 2007-01-02, 01:34

PG TPS took over a month to find it's first 100 primes...but only took 12 days to find it's next 100 primes!

Congrats to Rytis and all those crunching PG TPS!!!

Now if it takes just 6 days to find the next 100 primes, we'll really be cooking!

But who cares about the single primes...let's find the twin!

Skligmund · 2007-01-03, 05:56

Updates on my hosts:

Athlon64 3700+ @ 2800 MHz
Athlon64 X2 3800+ @ 2550 MHz
Athlon64 3500+ @ 2640 MHz
Athlon64 3200+ @ 2492 MHz
AthlonXP 2100+ @ 2038 MHz
Pentium III 1000 MHz x 2 (Dual Processor)
Pentium III 1000 MHz (only on about 30% of the time)
K6-III+ @ 600 MHz
K6-2+ @ 400 MHz (laptop)

In the near future, I plan on replacing the K6-III+ with a Duron 1200 that was just given to me.

I can't run any more computers in my apartment, I already blew the circuit breaker once this week... (of course my 2 500 watt lights were on, my stereo (big), my TV (tube) and maybe a toaster, and the fridge.. and the other fridge. Oh, and all my computers.

I stole my parent's P4 2.80 for the weekend, it did pretty good.

pacionet · 2007-01-03, 16:11

Jmblazek did you send the 2700-3000M to Rytis ?

Please send me also the file (Unfortunately I deleted the file gribozavr sent me)

gribozavr · 2007-01-03, 16:19

pacionet, you can download the file from the same link:
http://www.sendspace.com/file/dkk1x2

You can download a file from sendspace as many times as you wish, there are no limits (actually there are, but for large files).

Rytis · 2007-01-04, 17:57

I have been investigating mismatching LLR doublecheck results and found out that all that only checksum (?) was different, for example:
2639731083*2^195000-1isnotprime.LLRRes64:BFB1E48AC80A10C9Time:141.540sec.
2639731083*2^195000-1isnotprime.LLRRes64:181B742227F46C90Time:122.504sec.

Well, they both note that it's not prime. But there could possibly be others who have one prime and one not. Does anyone know why the same LLR task has different checksums? Maybe it is safe to allow mismatching checksums?

gribozavr · 2007-01-04, 18:11

This "checksum" is called residue. It should be always the same. One of those two tests is incorrect.

Rytis · 2007-01-04, 18:32

I've switched to 100% doublecheck. There are a lot of such results.

jmblazek · 2007-01-04, 19:23

I ran the test 3 times:

2639731083*2^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 144.099 sec.
2639731083*2^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 144.779 sec.
2639731083*2^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 150.652 sec.

AMD Athlon 64 3400+

100% doublecheck...OUCH! Hopefully someone can shed some light on this. It would be good to know if this is random...or if it's coming from a specific machine or machines, etc.

If there's no nice answer to this, then double checking down to 0M will be necessary and you'll have to write an additional script to compare residues.

Also, we'll need twice as much computing power to reach 1G a week.

Maybe submit your question over at RS or to the general mersenne forum since they've been LLRing for some time. Let's hope there's a good explanation.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
BOINC	Unregistered	Information & Answers	6	2010-09-21 03:31
BOINC.BE	BATKrikke	Teams	2	2010-03-05 18:57
Boinc	Xentar	Sierpinski/Riesel Base 5	4	2009-04-25 10:26
BOINC?	masser	Sierpinski/Riesel Base 5	1	2009-02-09 01:10
BOINC	bebarce	Software	3	2005-12-15 18:35

2007-01-02, 01:34	#169
jmblazek Nov 2006 Earth 1000000₂ Posts	PG TPS took over a month to find it's first 100 primes...but only took 12 days to find it's next 100 primes! Congrats to Rytis and all those crunching PG TPS!!! Now if it takes just 6 days to find the next 100 primes, we'll really be cooking! But who cares about the single primes...let's find the twin!

2007-01-03, 05:56	#170
Skligmund Dec 2006 Anchorage, Alaska 116₈ Posts	Updates on my hosts: Athlon64 3700+ @ 2800 MHz Athlon64 X2 3800+ @ 2550 MHz Athlon64 3500+ @ 2640 MHz Athlon64 3200+ @ 2492 MHz AthlonXP 2100+ @ 2038 MHz Pentium III 1000 MHz x 2 (Dual Processor) Pentium III 1000 MHz (only on about 30% of the time) K6-III+ @ 600 MHz K6-2+ @ 400 MHz (laptop) In the near future, I plan on replacing the K6-III+ with a Duron 1200 that was just given to me. I can't run any more computers in my apartment, I already blew the circuit breaker once this week... (of course my 2 500 watt lights were on, my stereo (big), my TV (tube) and maybe a toaster, and the fridge.. and the other fridge. Oh, and all my computers. I stole my parent's P4 2.80 for the weekend, it did pretty good.

2007-01-03, 16:11	#171
pacionet Oct 2005 Italy 101010011₂ Posts	Jmblazek did you send the 2700-3000M to Rytis ? Please send me also the file (Unfortunately I deleted the file gribozavr sent me)

2007-01-03, 16:19	#172
gribozavr Mar 2005 Internet; Ukraine, Kiev 11·37 Posts	pacionet, you can download the file from the same link: http://www.sendspace.com/file/dkk1x2 You can download a file from sendspace as many times as you wish, there are no limits (actually there are, but for large files).

2007-01-04, 17:57	#173
Rytis Nov 2006 1011000₂ Posts	I have been investigating mismatching LLR doublecheck results and found out that all that only checksum (?) was different, for example: 26397310832^195000-1isnotprime.LLRRes64:BFB1E48AC80A10C9Time:141.540sec. 26397310832^195000-1isnotprime.LLRRes64:181B742227F46C90Time:122.504sec. Well, they both note that it's not prime. But there could possibly be others who have one prime and one not. Does anyone know why the same LLR task has different checksums? Maybe it is safe to allow mismatching checksums?

2007-01-04, 18:11	#174
gribozavr Mar 2005 Internet; Ukraine, Kiev 11·37 Posts	This "checksum" is called residue. It should be always the same. One of those two tests is incorrect.

2007-01-04, 18:32	#175
Rytis Nov 2006 88₁₀ Posts	I've switched to 100% doublecheck. There are a lot of such results.

2007-01-04, 19:23	#176
jmblazek Nov 2006 Earth 2⁶ Posts	I ran the test 3 times: 26397310832^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 144.099 sec. 26397310832^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 144.779 sec. 2639731083*2^195000-1 is not prime. LLR Res64: BFB1E48AC80A10C9 Time : 150.652 sec. AMD Athlon 64 3400+ 100% doublecheck...OUCH! Hopefully someone can shed some light on this. It would be good to know if this is random...or if it's coming from a specific machine or machines, etc. If there's no nice answer to this, then double checking down to 0M will be necessary and you'll have to write an additional script to compare residues. Also, we'll need twice as much computing power to reach 1G a week. Maybe submit your question over at RS or to the general mersenne forum since they've been LLRing for some time. Let's hope there's a good explanation.