Too few factors found in 600M to 620M

alpertron · 2020-11-26, 01:06

I've just looked at https://www.mersenne.ca/status/tf/0/1/2/60000 and I found that there should be some hardware error:

21192 factors in the range 600M-610M today in the 72-bit range: only 3 factors found.
3268 factors in the range 610M-620M today in the 72-bit range: only 2 factors found.

In the first case there should be about 300 factors (maybe 200, but not 3). In the second case there should be about 45 factors, not 2.

It appears that the GPU being used is not helping the project.

S485122 · 2020-11-26, 02:13

Quote:

Originally Posted by alpertron

I've just looked at https://www.mersenne.ca/status/tf/0/1/2/60000 and I found that there should be some hardware error:

21192 factors in the range 600M-610M today in the 72-bit range: only 3 factors found.
3268 factors in the range 610M-620M today in the 72-bit range: only 2 factors found.

In the first case there should be about 300 factors (maybe 200, but not 3). In the second case there should be about 45 factors, not 2.

It appears that the GPU being used is not helping the project.

Probably a case of submitting the results with factors found one day and the others the next. On the 24^th, 162 successful and 0 unsuccessful attempts where reported in the 605M range.

Jacob

Mark Rose · 2020-11-26, 02:25

Appears to be so. Looking over a week and the numbers look reasonable: https://www.mersenne.ca/status/tf/0/7/2/60000

alpertron · 2020-11-26, 02:28

Thanks. So this was a false alarm.

Madpoo · 2020-12-05, 04:39

Quote:

Originally Posted by Mark Rose

Appears to be so. Looking over a week and the numbers look reasonable: https://www.mersenne.ca/status/tf/0/7/2/60000

There are a few users who submit their factoring work in separate batches of "no factor" and "factor". No idea why, since it seems like extra work to filter them into the separate piles, but whatever.

So that does mean there are strange spikes in factors found. But the line does smooth out over longer periods of time, and that's where you'd look to try and spot weirdness.

pinhodecarlos · 2020-12-05, 08:59

SRBase uploads in batches, but not continuous since some assignments can have a delay by BOINC users, have seen a few with 45 days delay. Anyway, all ranges will get done eventually.

pinhodecarlos · 2020-12-23, 09:12

Just using this thread to give you guys some feedback why the increase of SRBase output. There’s an interesting fight for third place at the boinc formula overall standings. Final figures within 8 days.

pinhodecarlos · 2021-04-22, 20:38

To inform SRBase has been chosen for the next sprint challenge at formula boinc (23.04. 02:00 (UTC) - 26.04. 01:59 (UTC) ). This means you will see a spike on returned work which is now currently at 650M for 72-73 bits. ~500k wus or more expected to be completed.

pinhodecarlos · 2021-05-02, 06:00

From project:

Quote:

We have a successful run in TF which completed 1,345,106 tests in 3 days. This is around 20 times more than in normal state.

Uncwilly · 2021-05-02, 06:35

They have run through 3 bit levels in the 600M range in less than a year. Watching the Mersenne.ca 'video' is interesting. When they started, there was a lot of clean-up of straggler ranges. Then the steady march across bit levels. Next pass starting down in the 1xxM range will be neat to watch.

pinhodecarlos · 2021-05-02, 07:16

It would be great to see SRBase been chosen for one of the GPU Boinc Penthalon disciplines, that would bring 50x more power or more during more days and help out clean 72-73 bits section. Yep, also looking forward to seeing this project impact on next bits level.

2020-11-26, 01:06	#1
alpertron Aug 2002 Buenos Aires, Argentina 2×727 Posts	Too few factors found in 600M to 620M I've just looked at https://www.mersenne.ca/status/tf/0/1/2/60000 and I found that there should be some hardware error: 21192 factors in the range 600M-610M today in the 72-bit range: only 3 factors found. 3268 factors in the range 610M-620M today in the 72-bit range: only 2 factors found. In the first case there should be about 300 factors (maybe 200, but not 3). In the second case there should be about 45 factors, not 2. It appears that the GPU being used is not helping the project. Last fiddled with by alpertron on 2020-11-26 at 01:06

2021-05-02, 07:16	#11
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2×3²×281 Posts	It would be great to see SRBase been chosen for one of the GPU Boinc Penthalon disciplines, that would bring 50x more power or more during more days and help out clean 72-73 bits section. Yep, also looking forward to seeing this project impact on next bits level. Last fiddled with by pinhodecarlos on 2021-05-02 at 07:19

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Nearly all Factors up to 2^71 found between 2.86 G and 2.96 G	Kalli Hofmann	Marin's Mersenne-aries	14	2021-04-12 13:12
No factors found	aketilander	PrimeNet	9	2011-05-17 11:32
CPU Time and Factors Found	Rodrigo	Operation Billion Digits	8	2010-08-14 20:36
500M-600M Range	moo	LMH > 100M	15	2005-10-21 01:10
How to find factors I found with TF?	edorajh	PrimeNet	3	2004-10-01 19:16

2020-11-26, 02:25	#3
Mark Rose "/X\(‘-‘)/X\" Jan 2013 13×227 Posts	Appears to be so. Looking over a week and the numbers look reasonable: https://www.mersenne.ca/status/tf/0/7/2/60000

2020-11-26, 02:28	#4
alpertron Aug 2002 Buenos Aires, Argentina 10110101110₂ Posts	Thanks. So this was a false alarm.

2020-12-05, 08:59	#6
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2·3²·281 Posts	SRBase uploads in batches, but not continuous since some assignments can have a delay by BOINC users, have seen a few with 45 days delay. Anyway, all ranges will get done eventually.

2020-12-23, 09:12	#7
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 5058₁₀ Posts	Just using this thread to give you guys some feedback why the increase of SRBase output. There’s an interesting fight for third place at the boinc formula overall standings. Final figures within 8 days.

2021-04-22, 20:38	#8
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2×3²×281 Posts	To inform SRBase has been chosen for the next sprint challenge at formula boinc (23.04. 02:00 (UTC) - 26.04. 01:59 (UTC) ). This means you will see a spike on returned work which is now currently at 650M for 72-73 bits. ~500k wus or more expected to be completed.

2021-05-02, 06:35	#10
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 11×967 Posts	They have run through 3 bit levels in the 600M range in less than a year. Watching the Mersenne.ca 'video' is interesting. When they started, there was a lot of clean-up of straggler ranges. Then the steady march across bit levels. Next pass starting down in the 1xxM range will be neat to watch.