mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   NFS@Home (https://www.mersenneforum.org/forumdisplay.php?f=98)
-   -   2021 15e post processing reservations and results (https://www.mersenneforum.org/showthread.php?t=26366)

Xyzzy 2021-04-24 19:29

Please reserve 11p4_284 for us.

:mike:

Xyzzy 2021-04-24 22:21

30941_59m1 is in post-processing.

:mike:

Xyzzy 2021-04-25 00:01

9p2_566L is in post-processing.

:mike:

Xyzzy 2021-04-29 00:29

Is the output data from remdups4 useful info to post and/or include with the log?

Example:

[C]$ zcat 11p4_284.dat.gz | ../remdups4 10000 > msieve.dat
Found 340831486 unique, 123853643 duplicate, and 4176 bad relations.
Largest dimension used: 1209 of 10000
Average dimension used: 1040.2 of 10000[/C]

:mike:

Xyzzy 2021-05-01 12:32

Please reserve 4p3_1341M for us.

:mike:

swellman 2021-05-01 13:56

[QUOTE=Xyzzy;577174]Is the output data from remdups4 useful info to post and/or include with the log?

Example:

[C]$ zcat 11p4_284.dat.gz | ../remdups4 10000 > msieve.dat
Found 340831486 unique, 123853643 duplicate, and 4176 bad relations.
Largest dimension used: 1209 of 10000
Average dimension used: 1040.2 of 10000[/C]

:mike:[/QUOTE]

I say no - most/all of the interesting information is contained in the log files already. Just my two cents.

VBCurtis 2021-05-02 02:30

The data point not in the log file that remdups provides, and I think is useful to others / future, is uniques vs raw relations. I post that data on almost every factor report.

For instance, an intrepid data-gatherer might look at unique % of raw relations to try to see what the optimal starting Q value is; lower Q sieve faster but yield more duplicate rels.

Dimension data is not useful, IMO.

swellman 2021-05-02 02:54

[QUOTE=VBCurtis;577400]The data point not in the log file that remdups provides, and I think is useful to others / future, is uniques vs raw relations. I post that data on almost every factor report.

For instance, an intrepid data-gatherer might look at unique % of raw relations to try to see what the optimal starting Q value is; lower Q sieve faster but yield more duplicate rels.

Dimension data is not useful, IMO.[/QUOTE]

Raw vs unique relations is contained in the msieve logs.

If folks want to include the remdups4 log, go ahead.

VBCurtis 2021-05-02 14:34

How does msieve ever see the raw relations count, if you run remdups first? I mean, the whole point of remdups is to feed msieve only the uniques.

You're correct about that log *if* remdups is not run.

swellman 2021-05-02 18:13

[QUOTE=VBCurtis;577437]How does msieve ever see the raw relations count, if you run remdups first? I mean, the whole point of remdups is to feed msieve only the uniques.

You're correct about that log *if* remdups is not run.[/QUOTE]

A good point. I personally almost never use remdups, I just let msieve do its thing.

Bottom line: raw vs uniques is nice data to have.

Xyzzy 2021-05-02 21:14

Which approach is more [URL="https://english.stackexchange.com/questions/38945/what-is-wrong-with-the-word-performant#38948"]performant[/URL] overall?


All times are UTC. The time now is 13:52.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.