I downloaded a couple of the latest hadam3p_anz models on the 14th and on just checking, found the following :-

1. Elapsed time is increasing at the rate of 2hrs for every 12hrs real time
2. Time To Completion is increasing
3. No trickles sent

It looks as though I shall have to abort them. Anyone else found the same problem?

It's probably because you had a different type of model running before, which set the Task duration correction factor to a certain value, and now BOINC is totally lost as to what that number should be now, so it keeps guessing.

(Yeah, yeah, I'll be finished in a moment.
I need another 5 minutes.
Be with you shortly, just hang on.)

**********

Or it could be something else. But Aborting everything that doesn't seem right is not the way to let BOINC learn.
And there's no more work, and may not be for "quite a while".

Thanks Les. I'll just let it run.

I just had at look at your 2 ANZ, and they took close to 3 days to get to a zip.

On my 3.5 Gig Haswell, running Windows under Wine on Linux, they took 6 hours to get to the first zip.
Just had another look, and the next lot have been sitting there for half an hour, so they're now on their way.
Zips are about 14 Megs.

And there's no more work, and may not be for "quite a while".

If only every time you posted about not knowing when work would appear more did!

It's all very odd. Each task does a bit of work, stops, "To Completion" increases a bit, task resumes, "To completion" drops more than the increase, does a bit of work, stops etc. etc.

I may have too little memory having upped to 'El Capitan'.

Anyway, when these tasks finish I'll be putting in more memory, so will see the result on the next tasks.

Dave Roberts -

Just curious. On the BOINC main screen, if you go to Options -> Computing Preferences -

Do you have -

100% of CPU's checked?

100% of CPU Time checked?

Under "When to Suspend", everything NOT checked?

---

I am running a Haswell 5820K in Windows 10.

I have six hadam3p.anz's running. Five are numbered in the hadam3p.anz_xxx 199012_12_287 series.

They have been running for over seven days and are showing two plus days to go.

NONE OF THEM HAVE RECORDED A TRICKLE and the time to go drops at about a third of the computer time.

The sixth is a 290 series and is behaving normally and will complete in under three

days as did another with a 289 number.

Worth persevering with?

Don

The ANZ models should have graphics, which, while not as good as "in the good old days", should show what is happening to the "model time".
You'll have to watch for some time, but you should be able to tell if the hours and days are constantly increasing, or if they're in a loop.
Perhaps write day the info at intervals.

Thanks Les

All five -289- series state"No model is currently running".

The remaining time on each has dropped about an hour over the last three hours.

It would appear that while they have about two and a half days to go it could take

another week or so at that rate and even then would you get a result?


The -289 - series example is running a model and trickles and is absolutely normal.

Hello Don

On the front page, right hand side is this: Script testing.
Perhaps you have one of these, in which case the advice is what we used to say 10 years ago: "Keep running them. Only the researchers know if they're good or bad."

Hopefully you'll be able to do this.

More, regular info would be nice. Then I could point the project people to this thread.

I've just emailed the project anyway.

I will keep running them to the end unless I am advised to the contrary.

Please advise what info you would like and I will do my best to oblige.

Ahh. Good question.
I usually play it by ear, and report things that are not going the way that the majority do.

Perhaps report what the graphics figures show re: progress. (Forward or looping.)

I also run with the 'net connection off, so that trickles and zips accumulate, so that I can see what's what. (When they get created, how big, etc.)
One thing that would show, (although it will also be in the Events Log/stdoutdae.txt), is if zips are being created and uploaded.

I should have checked first.
I've had a reply to my email:

Hi Les,

Thank you for this. Batch287 is the ANZ batch that was sent out previously and had a number of invalid combinations. My guess is that these runs relate to some of those setups. The invalid combinations have been corrected and resent in a new batch (batch 289).

Best wishes,
Sarah

So, dump batch 287.

Where's the info regarding the batch no.? I've looked at task and work details but can't see it.

Thanks in advance.

This is one of yours:

hadam3p_anz_q2g3_201212_12_290_010259670_2

Right hand column on the tasks tab in BOINC manager.