Primenet individual account report "current iteration"

[Peter Nelson]

I have been looking at my individual account report on primenet, and understand most of it.

After searching the forum for an answer, I'm still confused by the field "current iteration".

Under exponents assigned, some of my machines have been assigned queued up work to do next as well as the exponent they are currently working on.

The current iteration field contains a value against the exponents that are being worked on but not against the exponents not begun. Well, that makes sense, if you haven't started it shows blank.

However, I'm confused by the actual VALUE in this current iteration field.

My values are all one or two digit integers. eg 11,9,10,5
For some of these machines they have done many millions of iterations and I can't see any connection with the numbers reported nor factoring depth. Additionally there seems to be no recognisable significance in this field with whether the exponent is being Trial factored only or LL tested.

Please could someone explain what this field shows, and whether this can be interpreted usefully or is just an anomoly of primenet which used to be a feature but no longer applies.

Similarly, why are all my clients reported as V19/V20 when they are more like 23.8, 23.9 and 24.6? If it could show these correctly I would be able to better manage the rollout of upgraded client software onto my machines.
A possible workaround (which I dislike) would be to suffix my version number onto the end of my machine name and keep changing machine name when I upgrade software. eg MARS239 becomes MARS246 etc

Thanks in anticipation.

[Uncwilly]

The current iteration is the last that was reported to PrimeNet. It usually is low, unless you make the machine update. Until you communicate back to PrimeNet, it has no way of knowing your progress.

The version reporting has been stuck like that since I've been around. I think that we will have to wait until PrimeNet V5 to see that updated.

[Peter Nelson]

Hi Uncwilly,

Good suggestion but not right. I regularly use the manual communication and the update date/time stamp shows in the assignment report. I also have my machines set to check-in regularly even without the manual update. As some of these tests are a long way through their processing eg most % through LL, the figure 5, or 11 etc clearly cannot relate to the iteration number.

Still a mystery, so, what possibly can these low numbers mean?

As for version reporting, I suspect you may be right in that it will need a new server version before we see any change.

[PhilF]

Hello Peter.

Could it be that when the program is told to contact PrimeNet and update that it is simply sending the results file, but the results file has not been updated since early in the test? The default number of iterations before the results file is updated is 999999999. You might try lowering that number, and manually looking at the results.txt file to make sure the data is updated the way you want before using the manual communications option to send it.

Just a guess, I've never tried it...

-Phil

[patrik]

Could it be that you are doing trial factoring? (I.e. you have an F to the right of the exponent?) Trial factoring counts something I have forgotten, but your numbers look right then.

Otherwise check the "last updated" field. That should tell you if your computer has communicated.

[dave_0273]

If you are doing trial factoring... the "current itteration" would make sense. 1 means that it has only just started factoring up to 16ish which means that it is almost finished.

If you are doing LL testing, and you tell it to update while half way through the LL test, it should say what itteration you are up to. It has always worked for me....

[Peter Nelson]

Thanks for all your comments and suggestions.

OK, I've checked a few things. My main LL machine is processing but not online to primenet at the moment, hence no updating of the exponent.

My secondary LL machine I temporarily dedicated to trial factoring to pump up my factoring stats, and forgot about this when comparing.

My other machines are all doing TF.

This means that functionality is as described - for LL test, after updating primenet the report shows as expected. This solves part of my mystery.

Now what was still confusing me was then machines that were doing TF. Clearly the field is used to hold something else in this case, but what.

Someone said there was a table of values (I've observed values from 5 to 11) and their meaning for interpretation during factoring. Does anyone know where?

[cheesehead]

Originally Posted by Peter Nelson

Now what was still confusing me was then machines that were doing TF. Clearly the field is used to hold something else in this case, but what.

Someone said there was a table of values (I've observed values from 5 to 11) and their meaning for interpretation during factoring. Does anyone know where?

During TF, the "current iteration" field on the report shows which of 16 passes the TF was currently doing on its current bit level when the program reported progress. The possible range is either 0-15 or 1-16 -- I forget which.

For maximum speed, on each bit level TF tests factors in each possible congruence class mod 120 separately. That is, when testing factors between 2^63 and 2^64, TF first tries all factors that are congruent to 1 mod 120 (first pass), then all factors that are 7 mod 120 (second pass), then 17 mod 120 (third pass), and so on. (11 mod 120, 13 mod 120, etc. are skipped because a factor must also be +- 1 mod 8.) When the program reports progress to PrimeNet, it sends the TF pass number as the "current iteration".