mersenneforum.org  

Go Back   mersenneforum.org > Extra Stuff > Blogorrhea > kriesel

Reply
 
Thread Tools
Old 2019-07-17, 01:35   #34
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

132748 Posts
Default

Quote:
Originally Posted by hansl View Post
From new participant reference:

Needs a superscript: 2p-1
Code:
2[sup]p[/sup]-1
Oops. Done. Say hi to gretl.
https://www.youtube.com/watch?v=9246msCh7x4
kriesel is online now   Reply With Quote
Old 2019-07-17, 02:03   #35
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

22×3×5×97 Posts
Default

Quote:
Originally Posted by nomead View Post
About the manual assignment progress reports. Different programs have wildly different progress logs. Example from Mlucas:

[May 20 03:38:41] M50346019 Iter# = 24280000 [48.23% complete] clocks = 00:26:42.587 [160.2587 msec/iter] Res64: 54912F2731A45A41. AvgMaxErr = 0.042192089. MaxErr = 0.062500000. Residue shift count = 12144693.

But minimum common requirements should be current iteration count and Res64/shift so that interim residues could be stored, if needed.

I propose that instead of having the manual results submission form decipher many different log formats, there should be a simple common format, and let the client submission script do the conversion to that.
What I was looking for was manual progress reports that are submitted manually, by the user, not by some script. Motivations are to provide the server with interim residues from manual assignments, keep a manual assignment from being expired, and possibly detect error conditions earlier.

Perhaps a useful but easy form to enter would be:
<appname> <version> <console output line>
Only few <appname> <version> combinations would be supported.

Candidates are
CUDALucas v2.06
GpuOwL v6.5 and maybe v5.0 v3.8
These could be case-insensitve.

So an example would be
cudalucas 2.06 followed by an entire console output line copy/pasted. Like this:
cudalucas 2.06 | May 30 22:04:09 | M89000671 10000000 0xf2af4c9456abf0de | 5040K 0.09766 39.4100 197.05s | 36:01:03:19 11.23% |

Then the server can easily parse any such interim progress submission as
<appnamestring><spacedelimiter><appversionstring><spacedelimiter><consolelinestring>, parse the consolelinestring in that app/version's way, and get the exponent, iteration number, and interim residue, or if the appname version combination at the front is not a supported one, respond
"Application and version combination submitted is not supported by the manual results page."
If the error is because the user typed cudlucs instead of cudalucas, or 2.60 instead of 2.06, they can try again.

Last fiddled with by kriesel on 2019-07-17 at 18:15
kriesel is online now   Reply With Quote
Old 2019-07-22, 19:27   #36
ewmayer
2ω=0
 
ewmayer's Avatar
 
Sep 2002
Rep├║blica de California

266168 Posts
Default

Quote:
Originally Posted by kriesel View Post
Mlucas:
[Test|DoubleCheck]=[<AID>,]<exponent>,<how_far_factored>,<p-1_done>
DoubleCheck=B83D23BF447184F586470457AD1E03AF,22831811,66,1
Test=DDD21F2A0B252E499A9F9020E02FE232,48295213,69,0
Test=332220523,80,1
Mlucas can also handle just a prime exponent on a line by itself, which triggers an LL test of the corresponding Mersenne number. But if one is using the primenet.py script for assignment/results management, its "do I need to fetch new work?" calculation only uses official Primenet-server-style assignments, with the 32-digit hex code, like the first 3 lines in the above sample.

The next release, v19, will add PRP support.

Last fiddled with by ewmayer on 2019-07-22 at 19:27
ewmayer is offline   Reply With Quote
Old 2019-08-08, 15:35   #37
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

11128 Posts
Default

From point 5 of post 18 of the Prime GPU Computing reference material thread, you mentioned that mersenne.ca has exponent status up to 10^10. This is true, and it also has a list of exponents beyond that limit which have factors available to be downloaded (see the bottom of the page).
Dylan14 is offline   Reply With Quote
Old 2019-09-23, 06:29   #38
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
"name field"
Jun 2011
Thailand

34×112 Posts
Default

(grrr, man, my OCD gets ticked when I see #value and #div0 in those tables!)
LaurV is offline   Reply With Quote
Old 2019-10-13, 16:02   #39
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

26F316 Posts
Default Thanks for doing this!

Feel free to delete this post (as it isn't reference material), but I just wanted to say thanks for doing this.

The curation of this relatively complex subject domain is important and beneficial!


(Apparently referring to the Google Colab blog thread I've started in this blog; moved here from that thread; Kriesel)

Last fiddled with by kriesel on 2019-10-13 at 18:34
chalsall is online now   Reply With Quote
Old 2019-10-14, 16:44   #40
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

52·199 Posts
Default

How much faster is 5700 XT vs RX480? How long it takes to PRP on the 90M exponent range?
pinhodecarlos is offline   Reply With Quote
Old 2019-10-16, 01:12   #41
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

2·293 Posts
Default

I have built mlucas on the Colab, using the reverse tunnel code. See this and subsequent replies for debug stuff from me and Ernst.
And for post 9 of the Colab reference, another way to work with a worktodo.txt would be to directly connect to the server via SSH using the method described here and then use your favorite text editor to modify/create the file (which may require using apt-get to get certain ones, like emacs).
Dylan14 is offline   Reply With Quote
Old 2019-11-10, 21:49   #42
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

2·293 Posts
Default

A suggestion for the GIMPS progress thread: some people like to factor numbers, either partially or completely. For example, petrw1 has his thread on his efforts to get under 20M unfactored exponents below the PrimeNet boundary of 1G and GP2 routinely posts exponents which are fully factored. Perhaps these are things to track?
Dylan14 is offline   Reply With Quote
Old 2019-11-13, 02:15   #43
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

22×3×5×97 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
Perhaps these are things to track?
Thanks for your input. They occur to me as having little or nothing to do with finding new mersenne primes, the goal of GIMPS.
kriesel is online now   Reply With Quote
Old 2019-11-15, 23:02   #44
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

22·3·5·97 Posts
Default "Top of" navigation links addition

To hopefully aid navigation in this large and growing collection of reference posts, I've begun adding links for
Top of thread where I think it makes sense (page 2 or higher of multipage threads)
and Top of reference tree to reference posts, beginning with the gpu computing reference thread, which got
on pages 2 and 3, and
on page 1

Does this seem useful? Less clear than some other naming? Suggestions?
kriesel is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Mersenne Prime mostly-GPU Computing reference material kriesel kriesel 33 2021-10-24 23:43
P-1 discussion thread Rincewind Five or Bust - The Dual Sierpinski Problem 57 2011-02-06 21:53
Sieving discussion thread jasong Twin Prime Search 311 2010-10-22 18:41
PRP discussion thread philmoore Five or Bust - The Dual Sierpinski Problem 83 2010-09-25 10:20
Theological Discussion Thread clowns789 Soap Box 3 2006-03-09 04:05

All times are UTC. The time now is 18:54.


Thu Oct 28 18:54:33 UTC 2021 up 97 days, 13:23, 0 users, load averages: 0.91, 1.17, 1.23

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.