PrimeNet feature suggestions
 

Not sure how difficult it would be to implement these features, but I'd like to see:

(1) The ability to configure time zones on the PrimeNet pages, so that individual users (when logged in) could see things like current server time, completion times, assignment times, etc. in their native time zone, rather than in UTC as is displayed now.

(2) On the PrimeNet Registered CPUs page, the ability to hide (and unhide, as desired) inactive CPUs that a user is unable or unwilling to reconnect to GIMPS/PrimeNet. There exists, of course, the option of deleting such CPUs completely, but this (1) eliminates any historical record of such CPUs (e.g. GHz-days earned all-time, etc.) and (2) is not as amenable to reactivating such CPUs at a later date as maintaining a permanent record of such CPUs would be.

(3) On the Today's Numbers quotation, CPUs that have not reported *any* activity (i.e. have earned 0.000 GHz-days) in some reasonable period of time, say 365 days, should not be included in the count of Total CPUs. In fact, this should probably be ditto for users and teams (note that I am not saying these entities should be completely deleted from PrimeNet; they simply shouldn't be included in the Today's Numbers census). This would give a more accurate estimate of the true resources available to GIMPS, while excluding any accidental registrations, disconnected PCs, or attriters. Currently, PrimeNet reports almost 400,000 PCs, over 57,000 users, and over 400 teams. My guess is that a large percentage of these entities have become disconnected, or have quit GIMPS entirely. Perhaps hidden PCs in (2) should also be excluded from the Today's Numbers figures.

(4) On the Manual Testing page for Assignment Extensions, have the option to extend every assignment listed for 60 days with one click. There could be a check-box for "Select All Assignments" and if that is selected, when the Extend Assignments button is clicked, every single assignment to that user is extended for 60 days.

(5) Provide the ability to remotely transfer workunits between PCs. For instance, if Borg1 is disconnected from PrimeNet (and is irretrievable), allow the user to select an option to transfer some or all of Borg1's worktodo lines to other active systems. In other words, form the appropriate worktodo lines and transmit them to the other PC(s) as desired. This would be extremely helpful to folks administering borged systems remotely from hundreds or thousands of miles away.

(6) On the 1000 Most Recent Cleared Exponents report, allow the options to "Exclude TF" or "Exclude P-1" or "Exclude LL", etc. etc. a la a user's individual results page. There are times when heavy TFers submit their assignments near the top of the hour and essentially all of the last 1000 results are TF-LMH results. It might be interesting to be able to look back at the last 1000 doublechecks, last 1000 LLs, last 1000 P-1s, etc. etc. especially for the probability and statistics junkies among us.

(7) Add a countdown to completion of all first-time tests below M50,000,000 or even M60,000,000 (or both!) to the milestones page. Just for yucks, how about a countdown to M79,300,000 (a.k.a. the classical GIMPS upper limit)?

What do y'all think? Good ideas? Horribly misguided thoughts from beyond the grave? Impossible to implement without giving Snake{1,2,3} or Fish{1,2,3} cardiac infarctions?

I personally would enjoy some of those.

At least (3) and (5) have already been suggested:

[url]http://mersenneforum.org/showthread.php?t=14554[/url]
[url]http://mersenneforum.org/showthread.php?t=14520[/url]

[QUOTE=ixfd64;252139]At least (3) and (5) have already been suggested:

[URL]http://mersenneforum.org/showthread.php?t=14554[/URL]
[URL]http://mersenneforum.org/showthread.php?t=14520[/URL][/QUOTE]

This incarnation of (5) is a little beyond what I was thinking, though, as it requires transmission of the Pxxx and Qxxx save files to the PrimeNet server and then back out to the new borg. The problem with this is: What happens if someone has deleted everything from the old borg, or if the old borg is simply not accessible? PrimeNet isn't going to be able to magically restore the save files.

What I am proposing is a PrimeNet-side solution, where PrimeNet communicates with the borg of my choice and sends it the original worktodo line for the assignment being transferred.

I'd love to be able to send save files like was suggested in that other thread, but if I can't get to the borg (or if the Prime95 folder has been deleted), that feature isn't going to be very helpful.

[QUOTE=NBtarheel_33;252088]I'd like to see: [...] On the Today's Numbers quotation, CPUs that have not reported *any* activity (i.e. have earned 0.000 GHz-days) in some reasonable period of time, say 365 days, should not be included in the count of Total CPUs. In fact, this should probably be ditto for users and teams (note that I am not saying these entities should be completely deleted from PrimeNet; they simply shouldn't be included in the Today's Numbers census).[/QUOTE]

I don't know if the current system does this, but deleting inactive usernames is important. Without it you end up with new users not being able to get the username they desire because someone registered it years ago and never came back,

[QUOTE=NBtarheel_33;252088](7) Add a countdown to completion of all first-time tests below M50,000,000 or even M60,000,000 (or both!) to the milestones page. Just for yucks, how about a countdown to M79,300,000 (a.k.a. the classical GIMPS upper limit)?[/QUOTE]

You can indirectly and monthly see that here in this report I produce:
[url]http://www.mersenneforum.org/showthread.php?t=11292[/url]

[QUOTE=NBtarheel_33;252088](6) On the 1000 Most Recent Cleared Exponents report, allow the options to "Exclude TF" or "Exclude P-1" or "Exclude LL", etc. etc. a la a user's individual results page. There are times when heavy TFers submit their assignments near the top of the hour and essentially all of the last 1000 results are TF-LMH results. It might be interesting to be able to look back at the last 1000 doublechecks, last 1000 LLs, last 1000 P-1s, etc. etc. especially for the probability and statistics junkies among us.[/QUOTE]
This especially has now become more important for me. Any of these solutions would work for me (but listed in order of preference):[list=a][*]same as is now, but a longer report (>1000 entries), as long as each report covers the full 3600 seconds since the previous report[*]separate reports for each worktype (C|F|NF|NF-PM1|NF-ECM), as long as each report covers the full 3600 seconds since the previous report[*]either of the above, but with reports produced on a more frequent rotation (e.g. 15 minutes)[/list]The important point in any implementation is that not to miss any data. Even when there's not a large influx of manual TF results it's a rare thing that 1000 results spans back a full hour.

A more complex solution that could also be useful would be to have reports for the past 24h archived, either by worktype or all combined; on the update previous reports are shuffled down:
* [i]report_recent_results_23.txt[/i] is deleted
* [i]report_recent_results_00.txt[/i] -> [i]report_recent_results_01.txt[/i]
...
* [i]report_recent_results_22.txt[/i] -> [i]report_recent_results_23.txt[/i]


Is there any technical reason why something like any of the above couldn't be implemented?

[QUOTE=NBtarheel_33;252088](7) Add a countdown to completion of all first-time tests below M50,000,000 or even M60,000,000 (or both!) to the milestones page. Just for yucks, how about a countdown to M79,300,000 (a.k.a. the classical GIMPS upper limit)?[/QUOTE]From past experience, we know that there are a few folks here who (a) are impatient to see posted countdowns progress, (b) think it's a good idea to poach assignments of "slower" computers for the sake of "speeding up" progress toward milestones (ignoring that such poaching slows down progress toward higher milestones), and (c) have trouble controlling their impulses to poach.

I prefer not to see any increase in temptations for such folks.

If we're going to post additional countdowns, I recommend that they be posted somewhere that requires a more-than-minimal effort to see them. As an example, posting countdowns in a forum thread (such as petrw1's) rather than on an overly-convenient countdown summary page requires slightly more effort to access them, and thus presents less temptation to those who have trouble controlling the impulse to poach, or at least extends the time interval during which they can exercise such control.

petrw1's thread also shows that we can add a further layer of poach-discouraging indirectness by posting countdowns in attached files rather than directly in the thread.

[QUOTE=Uncwilly;252130]I personally would enjoy some of those.[/QUOTE]

[QUOTE=NBtarheel_33;252088](7) .... Just for yucks, how about a countdown to M79,300,000 (a.k.a. the classical GIMPS upper limit)?[/QUOTE]

[QUOTE=cheesehead;272569]From past experience, we know that there are a few folks here who .....

I prefer not to see any increase in temptations for such folks.

If we're going to post additional countdowns, I recommend that they be posted somewhere that requires a more-than-minimal effort to see them.[/QUOTE]
Just to see what it looks like at the moment:
[SPOILER]Of the 4,630,913 prime exponents in the range up to 79.3M:
2,901,722 (62.66%) have known factors,
615,659 (13.29%) have 2 matching LL's
491,405 (10.61%) have a single LL with a non-zero residue
622,093 (13.43%) are of unknown prime status
47 (0.000,95%) are confirmed Mersenne primes
[/SPOILER]

Warn me if Prime95 cannot write save files...
 

I have an i5-750 running all 4 cores on P-1. Since I have set MaxHighMemWorkers at 2 or 3 depending on the time of day, there will be times an exponent in Stage 2 will have to be put on hold in favor of another Stage 1 that uses less Memory.

Well, before a work unit is put on hold a save file is written...that is unless the hard drive is full...in which case it apparently just abandons the work WITHOUT WARNING and starts it over again when its turn comes up later.

Unless I just missed the warnings can I suggest Prime95 give an error message either in the Comm or Worker threads so I know a save file cannot be written and can react as I see fit.

I admit I was getting warnings from Vista that the hard drive was very full but it was NOT obvious to me it was affecting Prime95 so.

I estimate that for the month or so that this was happening I probably lost 20% of my thruput from missing save files that required work units to be restarted.

I am doing allot of TF work in the DC range, but grabbing work and releasing exponent is becoming a pain. I have to manually find wich exponent are factored enough (69 bit) out of the "Results Queries/Exponent Status" [url]http://mersenne.org/report_exponent/[/url] report, then remove them from my assignment without removing those at 67, and 68 bits... It's allot of copy pasting! and prone to errors. What would greatly help me, would be some additional column in the "My Account/Assignments" report [url]http://mersenne.org/workload/[/url]

If it could somehow show the Current Bit level and perhaps the B1/B2 range on an exponent, it would help alot of manual TFers

Many thanks!