Prime95 defaults - your input requested (round 2)
 

Right now, on initial install prime95 takes you through 4 dialog boxes:
1) Join GIMPS vs. Just-stress-testing
2) Primenet ID and computer name
3) Hours-per-day and memory settings.
4) Worker Windows - to change work preference
Then Primenet is contacted and assignments are made accordingly.

Under the theory that it is a good idea to force newbie's to get a DC assignment so that they can verify their hardware and get a feel for the time it takes work-units to complete,
I'm thinking of ditching the 4th dialog box. This would mean that work preference is left as what-makes-sense and the user will get a DC.

Good idea or not? Too cumbersome for experienced users that are setting up a new machine or re-installing prime95?

An experienced user can get around the DC assignment by changing the computer's work preference and unreserving the DC.

Another alternative would be to change the server to hand out a DC assignment to a new computer even if the work preference is set to LL or 100M. To me, this seems like a heavy-handed approach that would be very difficult the user to work-around if he insists.

Another idea is to ditch the 3rd dialog. The hours-per-day figure is somewhat useful (to the server in making assignment decisions and to the user in estimating completion dates) and unlikely to be filled in if the dialog box is removed from the initial dialog chain. The memory settings are almost irrelevant as LL assignments are almost always P-1'ed. Or maybe, move the hours-per-day to the Primenet ID dialog box.

Comments???

Until you dont touch worker windows , so ( as now) Prime can be used for PRP be free to make changes for LL and Primenet :)

There needs to be a way to access threads/workers preferences [b]before[/b] Prime95 goes ahead and reserves assignments.

Because until you put the primenet credentials, you cannot get to the individual worker settings. And once you ‘join Gimps’ Prime95 may quickly grab assignments before you had a chance to change configure the workers the way you want it (unless you pull the network cable, or something).

[QUOTE=Prime95;430058]Another alternative would be to change the server to hand out a DC assignment to a new computer even if the work preference is set to LL or 100M. To me, this seems like a heavy-handed approach that would be very difficult the user to work-around if he insists.[/QUOTE]

I think every new computer should do at least one DC so we get an early indication of the reliability of the computer instead of waiting five years to find out they produce bad results.

I don't think it's too much to ask.

[QUOTE=TObject;430063]There needs to be a way to access threads/workers preferences [b]before[/b] Prime95 goes ahead and reserves assignments.

Because until you put the primenet credentials, you cannot get to the individual worker settings. And once you ‘join Gimps’ Prime95 may quickly grab assignments before you had a chance to change configure the workers the way you want it (unless you pull the network cable, or something).[/QUOTE]

I have definitely had this happen to me before, sometimes even when I had already hand-loaded worktodo.txt with assignments. Worse yet, I have seen Prime95 simply dump and unreserve whatever is in worktodo.txt, fetch a bunch of new assignments, and begin work on those. By the time I could attempt to reclaim my original assignments, they had often already been assigned to another user.

[QUOTE=Mark Rose;430066]I think every new computer should do at least one DC so we get an early indication of the reliability of the computer instead of waiting five years to find out they produce bad results.

I don't think it's too much to ask.[/QUOTE]

With the chasm between the DC minimum and the first-LL minimum, it actually behooves the user to try a couple of DCs. It is quite possible that the fiftieth Mersenne prime to be discovered will be found as a double-check. On the other hand, the next Mersenne prime to be discovered on a first LL may have a nine-figure exponent and a discovery date ten years from now. I'd definitely be willing to take my chances on a double-check or two.

[QUOTE=Prime95;430058]Under the theory that it is a good idea to force newbie's to get a DC assignment so that they can verify their hardware and get a feel for the time it takes work-units to complete,
I'm thinking of ditching the 4th dialog box. This would mean that work preference is left as what-makes-sense and the user will get a DC.[/quote]
Don't do it! Every other piece of dialog is inconsequential, but not this one. This is the one choice that the user should be allowed to have.

[QUOTE=Prime95;430058]Another alternative would be to change the server to hand out a DC assignment to a new computer even if the work preference is set to LL or 100M. To me, this seems like a heavy-handed approach that would be very difficult the user to work-around if he insists.[/quote]
I think this is fine, except for 100M test. Every choice apart from 100M tells us that the user wants to find a prime and therefore it is good to establish the reliability of the machine by a forced DC. But 100M LL choice, to me, suggests user has a different focus -- they are going after the big prize. They probably won't like to be distracted with details like reliability or feasibility. Just git'er done. Besides, 1 DC is not going to establish the reliability sufficiently for a 100M test anyways.

[QUOTE=Prime95;430058]Another idea is to ditch the 3rd dialog. The hours-per-day figure is somewhat useful (to the server in making assignment decisions and to the user in estimating completion dates) and unlikely to be filled in if the dialog box is removed from the initial dialog chain. The memory settings are almost irrelevant as LL assignments are almost always P-1'ed. Or maybe, move the hours-per-day to the Primenet ID dialog box.
[/QUOTE]
Sounds good. It is probably good enough to assume 24x7 availability and use the rolling average to account for all the various extraneous factors.

----

One idea that I had (probably not feasible in the short term, but worth keeping in mind) is to create a cut-down version of P95 exclusively for GIMPS work. I mean, currently P95 is like a swiss army knife. A GIMPS-specific version would be just the knife, no other attachments. It would be tweaked for GIMPS (no stress testing code, no smaller unneeded FFTs, minimal dialogs, all cores for a single test, etc.). Probably a single screen with name, work preference, how many cores to use, and off you go.

[QUOTE=axn;430096]Don't do it! Every other piece of dialog is inconsequential, but not this one. This is the one choice that the user should be allowed to have..[/QUOTE]

Just to be clear, I am not proposing deleting this dialog box. I am proposing deleting it from the chain of dialog boxes prime95 walks the user through when he first installs and runs prime95. Thus, the new scheme would work as follows:

1) Install and run prime95.
2) Click "Join GIMPS" from the dialog box.
3) Enter your user id and computer name.
4) Computer contacts the server, gets a DC assignment and starts running.

From here you can go the menus and change the work preference, number of workers, unreserve the DC you were assigned, etc.

[QUOTE=Prime95;430097]Just to be clear, I am not proposing deleting this dialog box. [/QUOTE]
Understood. However, there will be some (a lot? most?) users who wouldn't explore the menus afterwards; they'd do the initial settings and that's that. What would be the work type for them? Would it be an initial DC followed by "whatever makes sense"? I suppose that'd be perfect for such users.

Here is a counter-proposal. Single screen:

* Primenet user id
* Worktype drop down (possibly don't show ECM, ECM-F, TF, P-1)
* Number of cores to use (presumptuous to use all cores by default, even though there is no real impact)
[Join Gimps button] [Cancel! I'm just stress testing button]

If they choose an LL work type, they get a 1st DC regardless.

Maybe have a "I want to ensure reliability of my machine by verifying a known result" checkbox, if we want to give the user the ability to opt-in/out of mandatory DC.

My thinking with the mandatory DC is that, with the (new proposed) default of using mutli-threading, a DC on a relatively modern machine should be a few days to a week, so it wouldn't be a huge wait for the user to get to the real action.

[QUOTE=TObject;430063]There needs to be a way to access threads/workers preferences [b]before[/b] Prime95 goes ahead and reserves assignments.

Because until you put the primenet credentials, you cannot get to the individual worker settings. And once you ‘join Gimps’ Prime95 may quickly grab assignments before you had a chance to change configure the workers the way you want it (unless you pull the network cable, or something).[/QUOTE]

I was thinking that too... like it could walk through the steps and skip over some of the things and leave them at defaults. But then instead of just going ahead and contacting the server when you're done with those simplified steps, show an "advanced setup" button to give people a chance to tweak those other settings if they really wanted to, *before* it contacts the server and gets work.

Or, since there are people out there who love to peek inside anything that says "advanced settings" even if they're unfamiliar with them, do something like exit the setup dialog and then have some big "start" and "stop" buttons on the GUI (instead of hidden under the Test menu). So it will NOT start right away, getting assignments, until they hit "Start". And the experienced users still have that chance to get in and modify other settings before it takes over.

I personally wouldn't mind having a stop/start button on the main GUI instead of in a menu, but then that's just me... I use those often when shuffling assignments around manually in the worktodo file. :smile:

EDIT: In other words, make the initial setup for new users as easy as possible, defaulting things to what would typically be the most reasonable choices... start out with a DC assignment no matter what because we want to prove the system is reliable, first and foremost, then "what makes sense" rules as they exist now. Too many dialogs for new users is just confusing and can be a barrier to getting started at all.

[QUOTE=Mark Rose;430066]I think every new computer should do at least one DC so we get an early indication of the reliability of the computer instead of waiting five years to find out they produce bad results.

I don't think it's too much to ask.[/QUOTE]

I'll ditto that. In my trawling through results of troublesome machines, I have been amazed on more than one occasion to find a system that NEVER did any DC work, and it's only years later when their own work is being checked by someone else that we realize nearly all of their results were bad.

It's a horrible shame those users were unaware of how bad their systems were at the time, it's a waste of resources, and it vastly increased the chances a prime was missed and won't be found until DC gets to it.

I mentioned in the strategic DC thread that I'm currently chasing down one particular computer that had several bad months in a row where I'm suspicious that most/all of their results over a 5 month period (perhaps longer?) are bad. And this is no lightweight system... they're turning in 30+ exponents a month in the 70M-80M range. And they're still active, and they're still not doing any DC work and may not even be aware of all the bad results we're turning up in their history.

Sigh... yeah, if Primenet were *forcing* (yes, that sounds bad, but...) them to do a DC once every 20-30 checks, I'd be hopeful that bad systems would be spotted sooner. It's only by chance that we're finding these systems now. I'm pretty sure the only reason this particular one popped onto my radar is that they got some exponents that were reassigned after being expired, and then the original person checked it in with a mismatch and then maybe that person saw the mismatch, decided to run it again and found that theirs was correct after all.

Otherwise I can't think of any particular reason why 3 out of 35 or so results that were turned in during a particular month would have happened to be double and triple-checked, proving this guy's results as bad, leading me to it. Sheer luck.

And who knows how many more like that are out there, just waiting to be discovered. There are thousands of systems out there that have never had a single one of their results double-checked, and that's actually something I've been trying to do here and there... do at least ONE double-check of each CPU that doesn't have one, but obviously it's not a one man job. :smile:

[B](note to George: maybe that could be a change to the DC assignments... hand those kinds of things out first, not just based on "lowest exponent first", but "most useful exponent first")[/B]

[QUOTE=Madpoo;430123][B](note to George: maybe that could be a change to the DC assignments... hand those kinds of things out first, not just based on "lowest exponent first", but "most useful exponent first")[/B][/QUOTE]

For the purpose of the first DC assignment, I would hand out CAT 4 assignments with only a single LL -- I would leave the the more useful exponents to users/machines that have a track recording of completely results quickly.

[QUOTE=Madpoo;430123]It's a horrible shame those users were unaware of how bad their systems were at the time, it's a waste of resources, and it vastly increased the chances a prime was missed and won't be found until DC gets to it.[/QUOTE]

This raises an idea...

Perhaps add some sort of data exchange between Primenet and the client which could then be presented via the Prime95's GUI as to the apparent health of the machine. Possibly with a link to a page on Primenet or MersenneForum along the lines of "What do I do if my machine is producing bad results".

I suspect there are many people who look at the local GUI much more often than logging into Primenet and drilling down far enough into the data to even realize their machine is producing suspect and/or bad results.

[QUOTE=axn;430106]
Maybe have a "I want to ensure reliability of my machine by verifying a known result" checkbox, if we want to give the user the ability to opt-in/out of mandatory DC.
.[/QUOTE]

Maybe reverse the sense of the wording?

Checkbox -> "Skip hardware verification assignment."

or similar.

To sum up the feedback:

1) My idea of altering the order and/or content of the initial dialog boxes wasn't great.
2) The server should assign a DC to new computers even if they select first-time LL or 100M LL. Madpoo desperately wants the data.
3) Perhaps an option either in the client or on the web pages could allow a user to skip these 2 DC assignments. GIMPS has always put the user in charge of what work their computer does, so this is probably a requirement -- we just don't need to make it particularly easy to evade initial DC assignments.

How about we generalize the DC rule --- let's call it the madpoo-needs-strategic-souble-checking-data rule:

Each and every computer must complete one verified LL test every year or one every 10 LL tests whichever is *less* frequent. There will be an opt-out (per-computer?, per-user?) accessed via the web pages. Perhaps, we should increase the requirements when a suspect result is reported.

This shouldn't be hard to implement. When a computer gets an assignment, we look for either a verified LL result within the last year or a currently active DC assignment. If none found, we assign a DC instead of a first-time LL test.

Comments?

Would that rule also apply to machines doing only ECM, TF, or P-1?
And if machines are not getting their assignments from PrimeNet, would the be 'force fed' a DC?

[QUOTE=Uncwilly;430205]Would that rule also apply to machines doing only ECM, TF, or P-1?[/quote]

No.

[quote]
And if machines are not getting their assignments from PrimeNet, would the be 'force fed' a DC?[/QUOTE]

I don't know how we'd do that, so no.

[QUOTE=Prime95;430204]How about we generalize the DC rule --- let's call it the madpoo-needs-strategic-souble-checking-data rule:

Each and every computer must complete one verified LL test every year or one every 10 LL tests whichever is *less* frequent. There will be an opt-out (per-computer?, per-user?) accessed via the web pages. Perhaps, we should increase the requirements when a suspect result is reported.

This shouldn't be hard to implement. When a computer gets an assignment, we look for either a verified LL result within the last year or a currently active DC assignment. If none found, we assign a DC instead of a first-time LL test.

Comments?[/QUOTE]
Does that rule be executed on all or on only new computers that start Prime95?
Second: in case all computers is affected can we disable that option editing prime.txt/local .txt

My proposal is all computers. It does not seem unreasonable to do a reliability check once a year....

You'd be able to override from a web page.

I have had memory go bad in computers on many occasions. Just because a PC has completed 17 double-checks it is not guarantee that the next test will be error-free.

Computers with ECC memory can self-report. Bad double-checks is how I get alerted to problems on non-ECC equipped machines (and justify running Prime 95 on company hardware, LOL).

[QUOTE=Prime95;430225]My proposal is all computers. It does not seem unreasonable to do a reliability check once a year....

You'd be able to override from a web page.[/QUOTE]

It's definitely the type of thing that should be ongoing. In fact, from what I've uncovered, many systems start out great and then as they head off into their sunset years, they begin to show increased instability.

It's been kind of interesting to take some of those test cases and track their bad/good progress over the years... the life and times of a modern CPU.

On the other hand there are systems that had some problem in the middle of their lifespan and then it was corrected. Maybe a bad mem module was replaced or they blew out the CPU fan.

Even more interesting are what I'd call "summer-itis" on some machines where they see to have more bad results in the warmer months (assuming they're northern hemisphere anyway).

It's easier to spot those trends when the system has a LOT of bad results, so I can't say it holds true in general, but point being, you could say an annual checkup is a good idea for computers and humans alike.

The yearly forced DC is implemented. The opt-out is on the web page [url]http://mersenne.org/thresholds[/url]

Let me know if I screwed anything up.

[QUOTE=Prime95;430423][B]The yearly forced DC is implemented.[/B] The opt-out is on the web page [URL]http://mersenne.org/thresholds[/URL]

Let me know if I screwed anything up.[/QUOTE]
Bravisimo! :tu:

[QUOTE=Prime95;430423]The yearly forced DC is implemented. The opt-out is on the web page [url]http://mersenne.org/thresholds[/url]

Let me know if I screwed anything up.[/QUOTE]I do approve the forced DC !

Since I only dabble in double-checks I tried that setting out. Nothing visible happens, in other words it is impossible to see if the yearly forced DC is implemented for my account or not. From the information displayed by the screen it is enforced no matter how often I click on the control button.

Jacob

[QUOTE=S485122;430428] From the information displayed by the screen it is enforced no matter how often I click on the control button.[/QUOTE]

Ah, but did you at least feel better having clicked the opt-out button??

Thanks, web page fixed.