mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Marin's Mersenne-aries (https://www.mersenneforum.org/forumdisplay.php?f=30)
-   -   Let's finish PRP-CF DC's. (https://www.mersenneforum.org/showthread.php?t=26165)

Uncwilly 2020-11-05 00:51

Let's finish PRP-CF DC's.
 
I am guessing that most of us doing PRP-CF and PRP-CF-DC are now using v30. If we started a push on the DC's I think we might be able to catch up the DC's to the first time check. Then when someone using v29 turns in a result, a DC will clear it. Other than that the certs will keep us current. I moved my one machine doing PRP-CF's to DC's.
Once we pass the next LL-DC milestone, I will move another worker onto the PRP-CF-DC's for a month or 2.

I don't know how to track the progress. James do you have any helpful suggestions for tracking this?

ATH 2020-11-05 02:32

I have 1 core on EC2 doing PRP-CF (v30.3b6) and 1 core doing PRP-CF-DC (v29.8b7).

Currently PRP-CF-DC is getting 9.23M exponent and PRP-CF is getting 10.62M.

I wish gpuowl could do these, then I could do a lot pretty quick.

pinhodecarlos 2020-11-05 07:53

I’m in.

kruoli 2020-11-06 11:24

Between 9,000,000 and 11,000,000, I counted 51,619 exponents that need a PRP-CF-DC (I also have an exact list of them, but I guess that's not as interesting). This number includes already assigned exponents. I might edit my script to exclude those in the future.

If wished for, I can run my script every week or so.

There are exponents way beyond 10,700,000 that already need a PRP-CF-DC. So we won't be finished as we reach the current PRP-CF first time test wave front.

I also set up five workers. Be aware that the current wave front's FFTs don't scale well on more than two or four threads. If you are interested in maximizing your throughput, do a throughput benchmark with FFT size 480K or so first.

pinhodecarlos 2020-11-06 11:37

How many unique ID workers do we have right now on the two ends?

kruoli 2020-11-06 13:13

According to mersenne.org, we have
[Code]_p95holder_
alan145
-Anonymous-
ANONYMOUS
arnaud
artorollo
Aseroberto
astraeus
ATH
bald2718281828
bpot
Bruno Olinto
bunddog
camaju
Carlos Pinho
Chad Steenerson
Chang Chia-Tche
C-Murray42
codest
Daniel Bamberger
darafshi
daveward2005
droseofc
ee33t1
Florian
fre_games
Glenn Raymond Hansen
Gordon Spence
happy5214
James Heinrich
Juergen Kampmeier
Justine V.
jvang
kgorski137
Klaus Nordmann
Kony
kracker
kutens
LaurV
Lord Julius
M Graber
M Miller
marcomich
Mariano
MarkS
marssystems
Massimo Cappelli (ik5pwa)
mat2022
MathDoggy
meracayn
Mikpos836
MindIsEssential
mnd9
mrh
Nesowa
Niels_Mache_Nextcloud
nordi
Oliver Kruse
oodaira
PageFault
Pham Cong Tai
Poao
Radish
riccardo uberti
RMJD
RobertKazan
roemer2201
Ruzovej
Sébastien Broucke
SebastianT.
Sergey Komolov
Silverfox
Stone
tanaydin
Team_Inspector
Todor
Tom
trebor
W4S0R1
XZT
Yuji Hasegawa
ZanK[/Code]
with PRP-CF-DC-assignments in the exponent range from 9,000,000 to 11,000,000.

pinhodecarlos 2020-11-06 13:29

Those are users, try by computer name just to see the difference from your list above, thank you.

kruoli 2020-11-06 13:34

Sorry, I do not know how to do this with the information on mersenne.org; if you know this, please share! :smile:

pinhodecarlos 2020-11-06 14:50

[QUOTE=kruoli;562423]Sorry, I do not know how to do this with the information on mersenne.org; if you know this, please share! :smile:[/QUOTE]

The only simple way I know is to go to recent cleaned results, copy paste onto an excel, apply conditions and filters. All my coding learning died after university years due to only not using it or not needed it for work. You know that joke where you learnt a lot of programming languages and softwares during your degree and then end up using excel on daily basis? That’s where I am now...

Uncwilly 2020-11-06 15:09

[QUOTE=kruoli;562402]There are exponents way beyond 10,700,000 that already need a PRP-CF-DC. So we won't be finished as we reach the current PRP-CF first time test wave front.[/QUOTE]But once we have caught the tail of the PRP-CF-FTC, then there should not be as much work to finish it off. The certs are reducing the future work load. Kruoli running your script once a week and posting the results will help us monitor our progress.:tu:

If we can temporarily shuffle more cores to the DC effort, that would help.
Anyone have a set-up like Chalsall does for P-1 on Colab for PRP-CF? He made it real simple: copy, paste, profit.

Runtime Error 2020-11-06 16:02

1 Attachment(s)
Just curious, is the effort in this thread responsible for the recent skyrocketing DC numbers? Wow!

[url]https://www.mersenne.org/primenet/graphs.php[/url]


All times are UTC. The time now is 10:43.

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