mersenneforum.org  

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

Reply
 
Thread Tools
Old 2019-11-12, 14:55   #1
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

2×3×37×43 Posts
Default 10 fold increase in new CPU's

According to the bottom graph we had a 10 fold increase in the number of new CPU's in the last 24-48 fours. It didn't jump that much in a long time. There is also a spike in new DC results.
Is there a rational reason for this?
Uncwilly is offline   Reply With Quote
Old 2019-11-12, 14:59   #2
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

2·307 Posts
Default

I bet it is the Colab/Kaggle thing. Every instance is a new CPU id.
PhilF is offline   Reply With Quote
Old 2019-11-12, 15:09   #3
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

23·3·193 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
According to the bottom graph we had a 10 fold increase in the number of new CPU's in the last 24-48 fours. It didn't jump that much in a long time. There is also a spike in new DC results.
Is there a rational reason for this?
Two big reasons:
Spike in DC is Ryan Propper.
Ben Delo has many hundreds of CPUs.
petrw1 is offline   Reply With Quote
Old 2019-11-12, 15:10   #4
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

2×2,543 Posts
Default

Quote:
Originally Posted by PhilF View Post
I bet it is the Colab/Kaggle thing. Every instance is a new CPU id.
Every 12 hour session is a new CPU id, which can really clutter up the user's GIMPS cpus report page,and also skew Primenet statistics in general as mentioned above. The fix is to put FixedHardwareUID=1 in the prime.txt file for the mprime configuration(s). Then on the user's cpus page, merge the multiple entries to the correct number of separate instances being run. I've run many sessions from two Google drive folders and 2 gmail accounts, and my GIMPS cpus page shows two colab cpus. GIMPS does not have a way to count gpus; even though I report results mostly flagged with uniqu system/gpu identifiers, they are all lumped into one manual results pile along with any manually reported cpu-performed work.
kriesel is offline   Reply With Quote
Old 2019-11-12, 18:27   #5
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

2×7×223 Posts
Default

If you use one of the newest versions of mprime and use the same local.txt for every instance and add this line to prime.txt:
FixedHardwareUID=1

then it will not create new CPUs for each session.
ATH is online now   Reply With Quote
Old 2019-11-12, 22:02   #6
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

2×3×37×43 Posts
Default

Quote:
Originally Posted by petrw1 View Post
Ben Delo has many hundreds of CPUs.
But Ben has been running for weeks now. This is an over night spike.
Uncwilly is offline   Reply With Quote
Old 2019-11-12, 23:57   #7
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

957010 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
But Ben has been running for weeks now. This is an over night spike.
Hmmm... Prime95 and Madpoo and James would be able to speak to this in more depth, but let me take a stab at it from an outsider's perspective, viewing only the Primenet Graphs page.

1. The number of new accounts doesn't really mean all that much, other than some bookkeeping by Primenet.

2. The "New results per day" (second graph) is a *big* deal.

3. Currently, the DCTF has effectively been done.

3.1. Hundreds of thousands of candidates ready to be assigned.

4. If this kind of sustained throughput is seen in LL'ing, there is a good chance many candidates will be assigned before being "optimally" prepared by first TF'ing and then P-1'ing.

4.1. This is not the end of the world.

4.1.1. Sub-optimal, in an ideal world. But we rarely find ourselves in such places.

4.2. Manage the situation.
chalsall is offline   Reply With Quote
Old 2019-11-13, 03:24   #8
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

7·1,069 Posts
Default

Sorry it has taken me all day to look at this.

Chris, please check your scripts. The new CPUs are from the GPU factoring accout, computer name ll_work.
Prime95 is offline   Reply With Quote
Old 2019-11-13, 06:56   #9
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

3·1,637 Posts
Default

For when a PRP DC rally with dedicated stats to narrow down the gap to PRP-WR?
pinhodecarlos is online now   Reply With Quote
Old 2019-11-13, 16:23   #10
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

256216 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Chris, please check your scripts. The new CPUs are from the GPU factoring accout, computer name ll_work.
Ah, shoot... Sorry George (et al)!

A slaved mprime used to update the status of "held" candidates was copying over the local.txt file from a version without the last updated date, but it also held the old GUID (from the server being retired). So, every time it was run a new GUID was generated...

I've shut that script down (since 0330 UTC this morning), and will apply the "Don't Generate New GUID" flag before running it again.

George and/or Aaron, please let me know if any of my other scripts are doing anything else stupid. They'll all be coming from the same (new) IP.
chalsall is offline   Reply With Quote
Old 2019-11-13, 18:18   #11
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

748310 Posts
Default

Quote:
Originally Posted by chalsall View Post
Ah, shoot... Sorry George (et al)!
No big deal. We just need to figure out how to clean it up with some simple SQL.

There are currently 54000 distinct CPUs apparently each controlling one assignment. I presume we want only one CPU controlling all 54000 assignments. If you PM me with the one and only ll_work computer GUID, I can pretty easily make the change and delete the spurious CPUs.
Prime95 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
CPU usage is only 50%: how to increase? Marsupial7 Information & Answers 18 2019-04-17 20:25
there is a way to increase the CPU usage? v44n Information & Answers 7 2016-07-07 06:08
Increase in RAM requirement soon? Jwb52z PrimeNet 4 2013-05-24 09:42
Will participation increase again? wouter Lounge 7 2005-02-26 21:50
How do they increase processor speed? clowns789 Lounge 17 2004-02-15 00:31

All times are UTC. The time now is 06:30.

Thu May 6 06:30:18 UTC 2021 up 28 days, 1:11, 0 users, load averages: 2.32, 2.39, 2.35

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.