PDA

View Full Version : A couple comments about pminus1 in general



jasong
07-03-2006, 07:42 PM
First, let me say, the "bosses"(What else could I call you?) are doing a bangup job of running the project. I'm just putting the following down as things to consider.

--The next.txt thingie needs to be fixed badly. Also, if it's fixed, I think there should be some general guidelines printed so people can be confident that they're not going to fall behind in their p-1ing.

One other thing, a lot of people may not be running p-1 because it makes their computer thrash when they try other things, or they're afraid it will interfere with their user experience in general. It would be nice if there were a thread to help people schedule Prime95 for specific times, as well as helping them deal with other DC projects, in terms of descheduling them and rescheduling them relative to pminus1.

Just some things to consider.

hhh
07-05-2006, 04:47 AM
I guess you know this already, just to put it down: the next.txt file doesn't work because the firstpass queue is empty. It ran empty because of an amok client requesting all the work for himself. The tests are now in the dropped test queue, hence the problem. It will fix automattically as soon as Firstpass reaches 12M (in a couple of months).
(On the other hand, if one could fix it, it would be better. But the Paladins of the forum would have done it already, right? Dunno.)
As for CPU and memory usage, there is this menu about daytime and nighttime settings; you can ajust it as you want. Do you need more accurate adjustment? Then you can put multiple exact times in the prime.ini, I think. I read it somewhere, probably in the readme.txt. I'm not sure anymore and too lazy to download it and check.
Wonder if I was of any help, yours H.

jasong
07-05-2006, 07:18 PM
I guess you know this already, just to put it down: the next.txt file doesn't work because the firstpass queue is empty. It ran empty because of an amok client requesting all the work for himself. The tests are now in the dropped test queue, hence the problem. It will fix automattically as soon as Firstpass reaches 12M (in a couple of months).
Sorry, I didn't know that. Kinda sucks, though, that it won't work for a couple months.

As for CPU and memory usage, there is this menu about daytime and nighttime settings; you can ajust it as you want. Do you need more accurate adjustment? Then you can put multiple exact times in the prime.ini, I think. I read it somewhere, probably in the readme.txt. I'm not sure anymore and too lazy to download it and check.
:haddock: I apologize, guys, I' m not too familiar with Prime95, so didn't realize that it already had that. Does it just affect memory, or can you schedule it to start and stop at certain times, to?

Wonder if I was of any help, yours H.
You were of tremendous help. Guess I should pay better attention before I post.

KWSN_Dagger
07-06-2006, 04:22 AM
Taken from the Readme.txt file.

You can configure this program to have different properties at different
times of the day and/or to not run during certain times of the day.
Unfortunately, you must manual edit the prime.ini file. Let's say you
want to install the program on a friend's machine and he runs a screen
saver at night. He also runs a disk defragmenter at midnight on weekdays.
This prime.ini file will run the program at a higher priority than his
screen saver at night and on weekends. It also sleeps for an hour when
his defragmenter starts running. Finally, at night fewer save files are
generated to allow his disk to stay powered down longer.

UserID=foo
Other prime.ini entries
Time=1-5/8:30-17:30
Priority=1
DiskWriteTime=30
Time=1-5/1:00-8:30,1-5/17:30-24:00,6-7/0:00-24:00
Priority=5
DiskWriteTime=240

The 1-5 refers to days of the week, with Monday=1 and Sunday=7. The
time portion refers to the hours of the day based on a 24-hour clock.
You do not need to specify days of the week (e.g. Time=1-7/0:00-8:00
is the same as Time=0:00-8:00). Unpredictable results will occur if
there are overlapping time intervals. Also note that any options that
appear in the Time= sections should not appear earlier in the prime.ini
file and you can no longer edit these options from the user interface.


SETTING AVAILABLE MEMORY
------------------------

The P-1 factoring step prior to running a Lucas-Lehmer test is more
effective if it is given more memory to work with. However, if you let
the program use too much memory then the performance of ALL programs will
suffer. The good news is that 98% of the time the program uses less
than 8MB. In fact, the program will work just fine if you instruct the
program to use only 8MB or less.

So how do you intelligently choose the available memory settings? Below
are some steps you might take to figure this out:

1) Be conservative. It is better to set the available memory too low
than too high. Setting the value too high can cause thrashing which
slows down all programs. Remember, the program will only use the
extra memory in stage 2 of P-1 factoring (about 12 hours a month).

2) Start with how much memory is installed in your machine. Allow a
reasonable amount of memory for the OS and whatever background tasks
you run (say 24 or 32MB). This represents the maximum value you should use.
The program won't let you enter more than 90% of installed memory.

3) Assuming you run your machine 24 hours a day, what hours of the
day do you not use your computer? Make these your nighttime hours and
let the program use a lot of memory during these hours. But reduce this
value if you also run batch jobs at night.

4) Factor in the information below about minimum, reasonable, and
desirable memory amounts for some sample exponents.

Exponent Minimum Reasonable Desirable
-------- ------- ---------- ---------
6000000 12MB 23MB 33MB
10000000 19MB 36MB 53MB
33000000 65MB 125MB 185MB

For example, my machine is a dual-processor with 128MB of memory.
Each CPU runs an LL test on an exponent near 10 million. On the off
chance that both CPUs wind up doing P-1 factoring at the same time,
I don't want to set the available memory to more than half of the
128MB. I guess Windows NT can survive on 24MB of memory. Thus, I set
the available memory to (128 - 24) / 2 or 52MB. This is my nighttime setting.
During the day, I set the available memory to 24MB. I can always stop
prime95 if it is doing P-1 factoring and I detect memory thrashing. More
casual users will probably want to set the daytime memory to 8MB so they
don't have to worry about it.

If at all in doubt, leave the settings at 8MB. The worst that will
happen is you end up running a Lucas-Lehmer primality test when stage 2
of P-1 factoring would have found a factor.

dudlio
12-15-2006, 09:46 PM
>First, let me say, the "bosses"(What else could I call you?) are doing a bangup job of running the project.

They're really not. They haven't found a prime in over a year and they've never allocated more than 20 users to sieving. User participation has been stagnant @ 1000 users for several years and forum participation has dropped to almost zero.

Don't get me wrong, SoB is a great project. But I think you'll find that the "bosses" gave up on the project a long time ago. It's been a user-run project ever since sieving was invented 2-3 years ago.

hhh
12-16-2006, 06:49 AM
They're really not. They haven't found a prime in over a year and they've never allocated more than 20 users to sieving. User participation has been stagnant @ 1000 users for several years and forum participation has dropped to almost zero.

Don't get me wrong, SoB is a great project. But I think you'll find that the "bosses" gave up on the project a long time ago. It's been a user-run project ever since sieving was invented 2-3 years ago.

:bonk: This had to be done.

It's not like the admins choose when to find a prime. It's not like they allocate people to do whatsoever. And sieving hasn't been invented 3 years ago. The project has started in March 2002, and sieving has been done since the very beginning, if I'm not mistaken.

As for the workload.
Some people might know it, I chose to help SoB's little sister project PSP by cleaning up the forum, and stuff. Sieving isn't automated there, it's more manual, etc. I can do this little part only because I'm student and don't have children.
And don't believe that because it's automated here, there isn' any work anymore, and party all the time. The factor submission have to be put into the database and gapchecked. The work has to be backupped. Imagine the server goes donw for whatever reason and has to be set up again. How much work? Count some hours, or a day. And you can't choose when to do this work, because there are 1000 people waiting. Even if you have this only twice a year, you'll be not amused.
And they have full time jobs.

Well, I'll stop here.

dudlio, you said in another thread that you would be interested perhaps in writing a sieve wrapper. Many people would be happy to have one. It hasn't been written up to now because it's plenty of work. Somebody who should know it estimated one man-month of work.
If you want to do it, you can PM me what you have done, what you think has to be done, and we can make a plan together, four eyes see more than two. If you decide not to do so, nobody will blame you.

I agree that there are many flaws in this project, but there are reasons for this, and one cannot summarize the situation by saying "They are all morons".
They are humans, and the project is running, which is worth a lot.

dudlio
12-30-2006, 03:35 AM
Thanks for the encouragement, hhh. On Wednesday I sent you a spec for a sieve wrapper program. As of now I've written about 500 lines of code. Most of the major components of the sieve wrapper are finished and seem to be working.

With the components working, I am testing the main program body, to see if it loops properly (detects done ranges, uploads factors, and starts a new range).

After that, there's some cleanup work, and the program will be available for review.

Cheers :)