Hello everyone,

Thanks for your discussion on the other thread (https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=5007). Based on your input we have decided to turn off Separation this Tuesday. N-body will still be running, and we appreciate anyone who wants to help crunch work for that application! I am not able to give you an exact time estimate because it's not clear exactly how long it will take to properly shut things down; however, I expect it to be sometime around noon EDT (1600 UTD).

We would like to once again sincerely express our gratitude for all the help that you have given in crunching for this project. We have learned a lot from Separation, and it's all thanks to you. We hope to see you stick around and help us continue crunching Nbody work in the future!

Best,
Tom

Do you mean only the work generator will be turned off this coming Tues, Jun 20th or a complete shutdown where in progress tasks are aborted or cancelled by the system?

That's fast... Just a week's notice? And here I was thinking that suggesting the end of the month would be fast, and saw a post mentioning a 30-day notice that seemed more fair...

Ah well.

Do you mean only the work generator will be turned off this coming Tues, Jun 20th or a complete shutdown where in progress tasks are aborted or cancelled by the system?

Very true , as the work unit deadline is normaly 12 days

Do you need the Separation WU in progess or should we just abort them and do other work?

Thanks for the quick decision and notice. I much appreciate knowing sooner rather than later.
Unfortunately, given my beater machine, I’ll move on. Was a pleasure helping the project and I wish you all the success in your research.

Do you need the Separation WU in progess or should we just abort them and do other work?

finish what you have

Fier d'avoir contribué à ce beau projet " Separation ".

---

Do you need the Separation WU in progess or should we just abort them and do other work?

finish what you have

The real question is wether we shall ask for more work which we can finish until the Tuesday or shall we move to other projects instantly as we are just generating heat because the results are not needed anymore and will be discarded anyway.

But in gereral I agree, there should be some delay between shutting down the work generators and shutting down the other daemons, which actually could run "forever" as the server will be running nBody and not shut down as well.

---

Are results still needed or will they be discarded?

We have all the results that we need for the science, so it doesn't make a difference to us whether you continue crunching Separation WUs between now and Tuesday. I am not sure if Prof. Newberg wants to just turn off the WU generator on Tuesday or if she wants to turn off the entire application... I think she wants to do the second one. I will double check next time we talk.

I know it's just a few more days but wouldn't it make sense to turn it off now? Seems wasteful.

I know it's just a few more days but wouldn't it make sense to turn it off now? Seems wasteful.

Although I agree to some extent. For people that are trying to reach credit and other goals I believe it would be in bad taste for them to turn work creation off early since a particular date has been said of Tuesday in June 20

Thanks Tom for explaining and checking what is going to happen with results that are currently in progress when Tuesday June 20 comes

I know it's just a few more days but wouldn't it make sense to turn it off now? Seems wasteful.

Very wasteful, keeps GPUs away from other projects, which actually need them and even CPUs away from nBody (and/or other projects). People who want to reach some credit goals can do it with nBody, just at a slower rate. Setting Milkyway to NNT.

For people that are trying to reach credit and other goals I believe it would be in bad taste for them to turn work creation off early since a particular date has been said of Tuesday in June 20

Well, work creation should have been turned off simply when the job is done.

---

And what I'm assuming is "a lot" of electricity required to power the GPUs to do work that will have no scientific value.
Only 2 more days but still very wasteful.

I'm going to assume they "have enough data" from what they've got, but adding two more days of data won't harm anything and will add to what they work with. I'm leaving mine crunching until the server says otherwise. As far as the deadline's concerned, I can plough through the max amount handed out (300 tasks per GPU or 900 tasks per host) in a couple of hours, so the question of letting stuff come back in doesn't exist for me.

---

The above was double spaced between sentences, I apologise for the forum software ruining my post.

Yes, we will be turning the entire Separation application off tomorrow. We aren't going to turn off the Wu generator and then let WUs trickle back in for a couple weeks, because we aren't going to use those results for anything.

Will all submitted tasks be validated, and points given for work completed, or will any unvalidtaed tasks just be left in limbo?

because we aren't going to use those results for anything.

Then you shouldn't be creating and sending them out IMHO, just resends of inconclusive results.

---

Looks like the well for separation tasks just went dry.... was 2705 ready to send, now at 0. :(