GPU to 72 status...

[chalsall]

Originally Posted by LOBES

I'm curious as to why GPU72 gave me the following P-1 assignment?

Because 108.3M is an "outliner", and needs some TLC to get it below 2,000 unfactored candidates.

OCD... And, yeah... We don't get out much...

[Zhangrc]

Originally Posted by LOBES

Here is another one in my queue from the same batch.

https://www.mersenne.org/report_expo...8349477&full=1

Please use larger bounds, B1=1e6, B2=4e7 at least. 108.3M should be treated differently, because factoring helps both GIMPS and sub-2k project.
In other ranges that have already less than 2000 unfactored, you may use smaller bounds.

[LOBES]

Originally Posted by Zhangrc

Please use larger bounds, B1=1e6, B2=4e7 at least. 108.3M should be treated differently, because factoring helps both GIMPS and sub-2k project.
In other ranges that have already less than 2000 unfactored, you may use smaller bounds.

That PC as 8063 MB of RAM. I have set it to use 6144 MB day / 6144 MB night.

I let Prime95 (V30.7 build 9) choose the bounds, I don't touch them. The only thing I change is tests_saved = 1, since the primality test will be a PRP with GEC and proof generation.

[Zhangrc]

Originally Posted by LOBES

The only thing I change is tests_saved = 1, since the primality test will be a PRP with GEC and proof generation.

It usually makes sense, but 108.3M should be treated differently at current situations.
If you use all these small bounds for 108.3M, there will be 2030 unfactored exponents at least, thus P-1 have to be rerun with larger bounds for the 2k project, which means that you previous run worth virtually nothing.
You may change a range (110.3M and 111.2M seems to be very good choices with less than 1900 unfactored).

[petrw1]

Originally Posted by LOBES

That PC as 8063 MB of RAM. I have set it to use 6144 MB day / 6144 MB night.

I let Prime95 (V30.7 build 9) choose the bounds, I don't touch them. The only thing I change is tests_saved = 1, since the primality test will be a PRP with GEC and proof generation.

For the sake of getting this range under 2000 you probably need to leave tests_saved =2.

[DrobinsonPE]

Over the last two days I was assigned the following exponents by GPU72.

Factor=N/A,19588279,73,74
Factor=N/A,19591483,73,74
Factor=N/A,19588337,73,74

Today I turned in the first two and am currently 85% done with the third.

Unfortunately TheJudger turned in all three exponents yesterday.

Just to be clear because my post is factual, This is just factual. I do not know how TheJudger and I got assigned the same assignments and I am not upset that it happened. I just thought that Chris would want to know that somehow the exponents got assigned twice. (my wife tells me often that people think I am mad when I am just being factual)

[LordJulius]

Originally Posted by DrobinsonPE

Over the last two days I was assigned the following exponents by GPU72.

Factor=N/A,19588279,73,74
Factor=N/A,19591483,73,74
Factor=N/A,19588337,73,74

If your lines always had 'N/A' in them, you never got an AID, so they were never officially assigned to you.
If your worktodo lines were "Factor=195____" when you started them, then likely the server had already assigned the exponents and bounced them back to you with 'N/A' inserted.

[Mark Rose]

Originally Posted by LordJulius

If your lines always had 'N/A' in them, you never got an AID, so they were never officially assigned to you.
If your worktodo lines were "Factor=195____" when you started them, then likely the server had already assigned the exponents and bounced them back to you with 'N/A' inserted.

GPU72 doesn't give assignment ids. It holds on to the assignment ids and the normal process is for a manual result to "poach" GPU72.

[LordJulius]

Originally Posted by Mark Rose

GPU72 doesn't give assignment ids. It holds on to the assignment ids and the normal process is for a manual result to "poach" GPU72.

And the Prime95 server can't know what GPU72 has reserved? Sounds like setting up a bunch of unintentional poaching then.

[chalsall]

Originally Posted by LordJulius

And the Prime95 server can't know what GPU72 has reserved? Sounds like setting up a bunch of unintentional poaching then.

Sorry, everyone... This was a profoundly Stupid Programmer Error on my part...

There was a legacy conditional from several years ago which resulted in Oliver being given a bunch of insane assignments. Oliver brought this to my attention, and I finally had some cycles to drill down, and figure out where the SPE was.

Again... Sorry about that everyone. A cycle is an awful thing to waste.

Everything /should/ be sane moving forward. Please report if things don't seem sane.

[Mark Rose]

Looks like I did ten unneeded assignments as well. A day of processing. Not the worst; oh well.