mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Data > Marin's Mersenne-aries

Reply
 
Thread Tools
Old 2020-07-22, 21:36   #1
Runtime Error
 
Sep 2017
USA

14310 Posts
Default PRP-VDF and the impact on DC's

Now that VDF-proof PRPs are on the horizon, how will that affect the LL-DC effort? Since LL-DC can result in the need for triple checks, that means that abandoning LL-DC in favor of VDF-PRPs might be optimal. Is that the future?

[Note: This was spun off from the Strategic DC thread.]

Last fiddled with by Uncwilly on 2020-07-24 at 03:31 Reason: Removed out of context material.
Runtime Error is offline   Reply With Quote
Old 2020-07-22, 21:55   #2
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

2·4,297 Posts
Default

Quote:
Originally Posted by Runtime Error View Post
Now that VDF-proof PRPs are on the horizon, how will that affect the LL-DC effort? Since LL-DC can result in the need for triple checks, that means that abandoning LL-DC in favor of VDF-PRPs might be optimal. Is that the future?
As soon as it is out of beta, doing PRP with VDF should be the first choice (if you can handle the temp storage and upload)
Doing the verification runs if you can handle the download, but not the storage) would be next.
DC LL's next (to clean up the tail end)
DC PRP's when the tail end hits that section.

Our special LL DC's and TC's will still have a place for a while.
Uncwilly is offline   Reply With Quote
Old 2020-07-24, 00:51   #3
Runtime Error
 
Sep 2017
USA

11·13 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
As soon as it is out of beta, doing PRP with VDF should be the first choice (if you can handle the temp storage and upload)
Doing the verification runs if you can handle the download, but not the storage) would be next.
DC LL's next (to clean up the tail end)
DC PRP's when the tail end hits that section.

Our special LL DC's and TC's will still have a place for a while.
Gotcha. And please forgive me because I know this is not a discussion thread, but just to confirm:

Suppose I want to double check a candidate that one person already found composite with LL. There is about a 2% chance that my LL residue won't match (is that estimate correct?). Since VDF is less than 1% of the effort, it is "cheaper" to ignore the previous LL result and just run PRP+VDF. Preferably, all unconfirmed LL results would be redone as PRP+VDF, as long as the double checker has the disk space and upload capacity. Does that sound right?

A downside to this is not discovering consistently bad machines or bad-faith actors.
Runtime Error is offline   Reply With Quote
Old 2020-07-24, 01:22   #4
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3×2,383 Posts
Default

Quote:
Originally Posted by Runtime Error View Post
A downside to this is not discovering consistently bad machines or bad-faith actors.
Another downside is we won't have work to give to machines that don't upgrade.

I'm seriously considering forcing all machines that do not upgrade (after a reasonable time period) to version 30 to do LL-DC work.
Prime95 is offline   Reply With Quote
Old 2020-07-24, 03:06   #5
Fan Ming
 
Oct 2019

1378 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Another downside is we won't have work to give to machines that don't upgrade.

I'm seriously considering forcing all machines that do not upgrade (after a reasonable time period) to version 30 to do LL-DC work.
It would be good to using small-size patches to upgrade automatically (if the machine didn't respond manually for a period time, that can be seen as the user agree to update by default), and stop assigning LL first test work from server(though still preserve LL test function in mprime software). With proper explanation of how strong the security of GEC (or VDF PRP-proof), and how low possibility that a number passed the PRP test is not prime (for example: if you found a mersenne number that passed the PRP test but not prime, then that means you found a thing that is much more awesome than a mersenne prime), I think PRP test is acceptable to most users.

Last fiddled with by Fan Ming on 2020-07-24 at 03:10
Fan Ming is offline   Reply With Quote
Old 2020-07-24, 03:34   #6
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

206228 Posts
Default

Quote:
Originally Posted by Fan Ming View Post
It would be good to using small-size patches to upgrade automatically (if the machine didn't respond manually for a period time, that can be seen as the user agree to update by default)
NO. No forced patches. It is my machine, I choose to d/l the software I want. This has been discussed elsewhere.
Uncwilly is offline   Reply With Quote
Old 2020-07-24, 03:59   #7
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

2·11·397 Posts
Default

I will still continue to do LL and LL/DC for a while. I have a bee under my bonnet about being (and staying) in the hundred percentile

Which is getting easier, with no competition, hehe, as all big guns switched already, or are on the way to switch, to PRP

Last fiddled with by LaurV on 2020-07-24 at 04:02
LaurV is offline   Reply With Quote
Old 2020-07-24, 05:41   #8
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3×2,383 Posts
Default

Guess we'll have to merge the PRP and LL into one Top Producers page.
Maybe offer double the CPU credit for a PRP test with a proof since it is worth twice as much to the project.
Prime95 is offline   Reply With Quote
Old 2020-07-24, 07:20   #9
Fan Ming
 
Oct 2019

5×19 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
NO. No forced patches. It is my machine, I choose to d/l the software I want. This has been discussed elsewhere.
[The thread is unlocked so I reposted here. Just delete the original post posted in another thread.]

That's never forced patches, but the alternative update. It's common sense that the machine was determined by the user, so I wrote that "if the machine didn't respond manually for a period time, that can be seen as the user agree to update by default", which means that the user can manually response "not to update".
Also, stop assigning LL first time test assignment doesn't mean stop accepting LL first time test results. If a user indeed want to submit first time LL results(most likely manual), that's of cource can be accepted.

Last fiddled with by Fan Ming on 2020-07-24 at 07:28
Fan Ming is offline   Reply With Quote
Old 2020-07-24, 07:25   #10
Fan Ming
 
Oct 2019

5·19 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Guess we'll have to merge the PRP and LL into one Top Producers page.
Maybe offer double the CPU credit for a PRP test with a proof since it is worth twice as much to the project.
I agree with that. And some measures need to be considered to deal with LL first test, like the idea I mentioned.
Fan Ming is offline   Reply With Quote
Old 2020-07-24, 08:17   #11
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

2×7×409 Posts
Default

Quote:
Originally Posted by Fan Ming View Post
"if the machine didn't respond manually for a period time, that can be seen as the user agree to update by default", which means that the user can manually response "not to update".
User gets to choose to update or not. You don't get to tell them to update. Regardless of how you want to word it in some eloquent fashion, it is just another way of saying "forced update".
retina is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Impact of AI xilman Lounge 19 2017-01-26 16:03
First pre-impact discovery for NEO search! cheesehead Astronomy 42 2013-11-22 04:54
GPUs impact on TF petrw1 GPU Computing 0 2013-01-06 03:23
GPU TF work and its impact on P-1 davieddy Lounge 161 2011-08-09 10:27
Another Impact on Jupiter Spherical Cow Astronomy 24 2009-08-12 19:32

All times are UTC. The time now is 09:03.

Sun Sep 20 09:03:56 UTC 2020 up 10 days, 6:14, 0 users, load averages: 1.92, 1.42, 1.41

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.