mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Data

Reply
 
Thread Tools
Old 2021-06-19, 03:53   #496
masser
 
masser's Avatar
 
Jul 2003
wear a mask

32148 Posts
Default

Quote:
Originally Posted by LaurV View Post

Any other people want to join the fry?
Ahem. You are still 5th to my 4th in the P+1 category.

Bring it on.
masser is online now   Reply With Quote
Old 2021-06-19, 11:08   #497
ZacHFX
 
ZacHFX's Avatar
 
Mar 2017
Halifax, NS

3×7 Posts
Default

Now this is the kind of thing I want to put my GPU to work on! If I'm understanding right, this range could use some TF to 70 bits? I didn't want to step on any toes, so I went way out of the way of ranges that looked like someone might be working on them. I'll start on 12.10, but if anyone wants to point me elsewhere, please do so!
ZacHFX is offline   Reply With Quote
Old 2021-06-19, 14:58   #498
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

13·192 Posts
Default

Quote:
Originally Posted by ZacHFX View Post
Now this is the kind of thing I want to put my GPU to work on! If I'm understanding right, this range could use some TF to 70 bits? I didn't want to step on any toes, so I went way out of the way of ranges that looked like someone might be working on them. I'll start on 12.10, but if anyone wants to point me elsewhere, please do so!
That should be fine. As long as people indicate here what they're working on, all toes will be spared.

Pretty much all incomplete ranges will need some TF and more P1.

Thanks
petrw1 is offline   Reply With Quote
Old 2021-06-19, 15:35   #499
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

13·743 Posts
Default

Quote:
Originally Posted by masser View Post
Ahem. You are still 5th to my 4th in the P+1 category.
Coming for you too!
Unfortunately, as long as P+1 is CPU only, you will probably run circles around me, haha
LaurV is online now   Reply With Quote
Old 2021-06-20, 00:20   #500
ZacHFX
 
ZacHFX's Avatar
 
Mar 2017
Halifax, NS

3×7 Posts
Default

Quote:
Originally Posted by petrw1 View Post
That should be fine. As long as people indicate here what they're working on, all toes will be spared.

Pretty much all incomplete ranges will need some TF and more P1.

Thanks
Great, thank you. It turns out I can crank through the entire 12.1X range up to 71 bits in about 10 days, so I'm already off to the races.
ZacHFX is offline   Reply With Quote
Old 2021-06-21, 02:36   #501
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

22·7·349 Posts
Default

Quote:
Originally Posted by ZacHFX View Post
It turns out I can crank through the entire 12.1X range up to 71 bits in about 10 days, so I'm already off to the races.
In the voice of Mr. Burns... Excellent...
chalsall is offline   Reply With Quote
Old 2021-06-24, 11:24   #502
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

965910 Posts
Default

Quote:
Originally Posted by masser View Post
Ahem. You are still 5th to my 4th in the P+1 category.
Bring it on.
Quote:
Originally Posted by LaurV View Post
Coming for you too!
Two with a single kick! You have to call me "mawashi geri" from now on!
Click image for larger version

Name:	mas1.JPG
Views:	26
Size:	12.8 KB
ID:	25182
Click image for larger version

Name:	mas2.JPG
Views:	23
Size:	12.5 KB
ID:	25183
(albeit this was accidental, I didn't go for it, it just happened that you were one position above in both yearly and lifetime tops, and my ECM workers were on full throttle; I switched them to ECM-F for now, to give you time to breath... )

Joking apart, we finished the 35.1M range, and found 71 factors from the 77 desired (mersenne.ca will catch up tomorrow, right now it only shows 68 factors), which took us 16 days. That makes 1 factor in 29.23 trials, considering the 2076 candidates we started with.

That last subrange (35.19M) didn't produce the number of factors we were counting (like 10 or 11, considering that is has more candidates there) so we are now under the target. Not mentioning that we got a really huge (composite) factor which could better have been two different factors and eliminate two exponents instead of only one exponent! What a waste! But all in all, our luck was quite good.

Now, for the next step, as axn suggested, we will try to go one bit higher (or deeper?) with TF, but we want to do that (for a start) only for the "bad" sub-ranges which still have 200 or more candidates left, maybe we get lucky and kick two rabbits with a single mawashi again - i.e. getting the subranges under 200 expos each, albeit this is quite impossible with the current "scores", but a single bit advance in 4 subranges (close to 900 candidates) would, in theory, give us the 6 more factors we miss. In fact, the theoretic score should be 8 to 9 factors by raising those 4 subranges to 75 bits.

Then: @Chris, is there any scenario possible in which your spiders grab the 35.1xM ranges from the server and pass them to my Colab, for x={0, 2, 7, 9}, or should I handle them manually? (which will take a lot more time).

If not, is there a scenario possible in which at least the reports can be done through gpu72 API? (adding all the assignments by hand to worktodo in colab is just a few minutes job, but it is reporting of the results periodically what makes the job boring...). I ask because I see that you still didn't decide to offer this possibility (few of my results that I reported to you but were not assigned through you are still stuck in that table).

Last fiddled with by LaurV on 2021-06-24 at 11:31
LaurV is online now   Reply With Quote
Old 2021-06-24, 11:48   #503
Viliam Furik
 
"Viliam Furík"
Jul 2018
Martin, Slovakia

619 Posts
Default

Quote:
Originally Posted by LaurV View Post
If not, is there a scenario possible in which at least the reports can be done through gpu72 API? (adding all the assignments by hand to worktodo in colab is just a few minutes job, but it is reporting of the results periodically what makes the job boring...). I ask because I see that you still didn't decide to offer this possibility (few of my results that I reported to you but were not assigned through you are still stuck in that table).
You can do it in under a minute, using the worktodo generating tool, originally meant for additional factoring, hence its name. It can be found here.

Setting the number of known factors to 0 makes it generate TF, P-1 and P+1 assignments for a given range and a given B1 for unfactored exponents.
Viliam Furik is offline   Reply With Quote
Old 2021-06-24, 15:09   #504
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

977210 Posts
Default

Quote:
Originally Posted by LaurV View Post
Then: @Chris, is there any scenario possible in which your spiders grab the 35.1xM ranges from the server and pass them to my Colab, for x={0, 2, 7, 9}, or should I handle them manually? (which will take a lot more time).
I can facilitate this. Not that much work on my side.

Could you please activate one of your Colab instances? It will initially be helping Wayne in 28.6M to 73.

Somewhat ironically, I no longer have any access to Colab GPUs. Or if I do get a GPU every week or so, it lasts about ten minutes. This makes testing somewhat problematic; I do have a "simulated" Colab environment, but it's currently on another job.

Lastly, so there's no "toe stepping", I don't actually reserve the candidates from Primenet for the "Less than 2K" sub-project. I'll start with 35.1M (to 75), and we can expand to other ranges as Wayne et al decide "makes sense".

Thoughts?
chalsall is offline   Reply With Quote
Old 2021-06-24, 15:24   #505
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

13·192 Posts
Default

Quote:
Originally Posted by chalsall View Post
II'll start with 35.1M (to 75), and we can expand to other ranges as Wayne et al decide "makes sense".

Thoughts?
Based on my personal P1 vs TF balance if all remaining 3xM ranges get TF to 75 I can complete them with P1....except maybe not 35.3. In this scenario which is done first doesn't matter.

If on the other hand Mr. L or anyone else has designs on more super aggressive P1 especially 35.3 then the remaining 3xM will only need partial TF to 75. In this scenario P1 should be first.

Either way I expect 3xM to be complete in 2021.

Then I'll take a deeper look at 2xM. This range will take a lot longer.
But basically most subranges will need TF 73 and decent P1 effort.

Last fiddled with by petrw1 on 2021-06-24 at 15:28
petrw1 is offline   Reply With Quote
Old 2021-06-24, 15:28   #506
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

22×7×349 Posts
Default

Quote:
Originally Posted by petrw1 View Post
If on the other hand Mr. L or anyone else has designs on more super aggressive P1 especially 35.3 then the remaining 3xM will only need partial TF to 75. In this scenario P1 should be first.
I'm just the facilitator here. Please tell me what to do.
chalsall is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Thinking of Joining GPU to 72 jschwar313 GPU to 72 3 2016-01-31 00:50
Thinking about lasieve5 Batalov Factoring 6 2011-12-27 22:40
Thinking about buying a panda jasong jasong 1 2008-11-11 09:43
Loud thinking on irregular primes devarajkandadai Math 4 2007-07-25 03:01
Question on unfactored numbers... WraithX GMP-ECM 1 2006-03-19 22:16

All times are UTC. The time now is 13:54.


Tue Aug 3 13:54:44 UTC 2021 up 11 days, 8:23, 1 user, load averages: 3.27, 3.49, 3.52

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.