Starting new bases

[gd_barnes]

Originally Posted by rogue

I'm pretty certain that I know the change that caused the memory leak. It was tied to the changes PFGW output to the GUI. I fixed that issue in the current release, but it is possible that other memory leaks exist. I have not had the time to investigate further. Which version did you experience the memory leak with?

When I have a chance, I'll run PFGW for a few hours (or days) and see if it is still leaking and if so, try to discover where the leak is. I'll most likely use base 3 because it will go through many more tests in that time frame than it would for most other bases. Can you tell me the upper bound of n that was being used when you ran into the problem? I don't expect that to be a factor, but you never know.

BTW, if I need to make any changes, I will probably modify PFGW to automatically apply -a1 to primality tests until that problem is resolved. If you have followed the LLR 3.8.0 threads it appears that it might also be affected by the same problem in gwnum.

Hum, you fixed it with the current release? Isn't the current release 3.3.0? That is what I was running. I suspect it also exists in 3.2.7. BTW, for base 3 on the GUI, I set "Output Verbose Screen" to "super quiet" if that tells you anything.

One thing that I noticed that I mentioned before just with 3.3.0: When the form "trivially factors prime", it displays that in the GUI when set on "screen normal" mode. That is irritating and it shouldn't do that. Previously it only displayed PRPs on the screen. Of course I don't use that for base 3 due to the millions of small PRPs but I wonder if that is a related issue to what you are talking about.

If you run base 3 for just a few MINUTES with the task manager open in Window's, you'll see the memory utilization slowly but consistently rising. You don't have to run it until the program stops. It took me over 30 mins. to figure out why my cores had stopped and I just checked memory usage on a whim. I thought at first that I had some endless loop in the script until I remembered this thread about memory leaks.


Gary

[rogue]

Originally Posted by gd_barnes

Hum, you fixed it with the current release? Isn't the current release 3.3.0? That is what I was running. I suspect it also exists in 3.2.7. BTW, for base 3 on the GUI, I set "Output Verbose Screen" to "super quiet" if that tells you anything.

One thing that I noticed that I mentioned before just with 3.3.0: When the form "trivially factors prime", it displays that in the GUI when set on "screen normal" mode. That is irritating and it shouldn't do that. Previously it only displayed PRPs on the screen. Of course I don't use that for base 3 due to the millions of small PRPs but I wonder if that is a related issue to what you are talking about.

If you run base 3 for just a few MINUTES with the task manager open in Window's, you'll see the memory utilization slowly but consistently rising. You don't have to run it until the program stops. It took me over 30 mins. to figure out why my cores had stopped and I just checked memory usage on a whim. I thought at first that I had some endless loop in the script until I remembered this thread about memory leaks.

No, 3.3.1 is current. Look here, http://www.mersenneforum.org/showpos...7&postcount=20. There is no 3.3.1 for Linux or Mac as 3.3.1 was Windows only. Although not directly in the release notes, the change it mentions in the release notes addressed a memory leak as part of the change.

Removing "trivially factors prime" is a bit more tricky. I can certain do it, but it was an existing feature of PFGW (before I took over the development). I'm wary of removing it because long time users of PFGW most likely still want that output displayed.

I will take some time on Saturday or Sunday to run a test to see if the leak was fixed in 3.3.1 or still exists.

[gd_barnes]

OK, I'll download 3.3.1. Thanks for the info. In looking at the timing of things, I see that 3.3.1 came out just after I left on my previous business trip.

The fact that it's only a Windows release is OK for now. It was my new Windows I7 that I've been double checking base 3.

Whew...tough keeping all of the machines up to date on this stuff.

[gd_barnes]

3.3.1 doesn't correct anything, the memory leak still exists, and it introduces additional problems, mainly on small tests. Therefore I recommend that people do not use it and that the version be completely deleted. IMHO, 3.3.0 should be used to make future changes.

Mark, please run a parallel test on WinPFGW 3.3.0 vs. 3.3.1 using the attached base 3 starting script per some details in the PM that I just sent you.

Here are some guidelines for the parallel test:

1. Set trial factoring to 30%, i.e. -f30.
2. Do not use the -l switch. Results aren't needed on such small tests.
3. Set "screen verbose" to super quiet.
4. Run both versions as close to exactly the same time as possible on at least a dual-core Window's machine with nothing else running in the background or on 2 Window's machines that run at exactly the same speed. This will demonstrate the speed and output difference.
5. Let it run for 15-30 mins. to demonstrate some things that I discuss in the PM.
6. Observe the memory usage in the task manager on both runs as it progresses and see that there is no significant difference. They both continue increasing.
7. Observe how easy it is to read the changing tests on the screen in 3.3.0, which gives you a clear status of where it is at at a glance. Now observe how you cannot read what is flashing by in 3.3.1 because it is displaying far more than it needs to, which also results in slowing the tests down by 1-2%; IMHO not a trivial decrease in speed. All of this despite the screen being in "super quiet" mode.

This is the kind of parallel testing that I keep talking about and harping on constantly when releasing new versions of things. You have to demonstrate that just because something was presumably fixed that nothing else was affected.


Gary

[MyDogBuster]

Parallel tested v4.3 on 6 bases. No problems. 2 of the bases had ck's over 15K and they were perfect.

[Batalov]

A script for algebraic elimination hints for pl_remain.txt
hiddenPowers.zip

If it reports
* x^3 ...
then it means the whole k can be eliminated (for a hypothetical example 8*125^n+1) -- practice shows that even squares are occasionally missed.

If it says
1|3 46^3 184*529^n+1
then just eliminate n=1 (mod 3) from the ...b529_k184... file