Hi,

I'm probably doing some amateur mistake here. Nevertheless:

>Requesting new tasks for CPU and NVIDIA GPU
>Scheduler request completed: got 0 new tasks

Even after repeated "Update" clicks, nothing happens. It connects but gets 0 new tasks.

CPU and NVIDIA GPU enabled, of course.

I have this problem on 3 out of 4 computers (for example: https://setiathome.berkeley.edu/show_host_detail.php?hostid=8322093 )

All 4 computers run other BOINC projects day and night. I just can't get SETI@home to run on them.

SETI@home works fine on my fourth computer.

There are many things that can affect how many tasks you get from a particular project.
Some of them:
Project has no work available at the exact time of request - all projects send out work in "lumps", so if someone has just grabbed the whole of a lump then :-(
The "resource share" for a project is too low for the computer to get work at the time of request - you get a different message, or no request is made. BOINC uses a long time effort share which often confuses people because they think a "50% share" on each of two projects will continuously get them equal amounts of work from them both.
The "get work" options aren't set correctly - for SETI you need to have the "AstroPulse", "Seti at home" and "other applications" boxes checked.
The drivers for the GPUs are incorrect - I notice you are running Windows 10 on the two computers than don't appear to have done any work. Sadly MS supplies drivers that are seriously crippled when it comes to doing computations - For nVdia GPUs only use the drivers supplied by nVidia.
A firewall is stopping BOINC doing communication properly
and so on......

(Others will no doubt come along and add to this list - many of us have tripped up on this sort of issue over the years)

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Thanks for your observations.

Turns out that what was needed was a couple of hours of patience - then 2 of the computers started getting work units. (Clicking "Update" in the meanwhile didn't help.)

A bit weird, I must say. (PrimeGrid, for instance, more or less forces work units onto my computer.) Not sure what to make of it.

Even so, I'm happy that it works now.

Thanks again!

Does PrimeGrid have a higher resource share or shorter deadlines? It may get preferential resources of your system if either or both answers are yes.

Pressing Update isn't going to have much effect after a period a while back were a number of heavy-hitters that were abusing this button and scripting this automatically to make sure their workload was always topped off at the expense of causing DDoSs to the rest of the users of the project. Now there is an automatic backoff period where if the last check was within the last X mintutes, pressing Update won't force communication or download of new work (where X is a random backoff time specified server-side).


* This was all true as of a few years ago unless they've changed the server/client code since then. Someone else can probably correct me if I'm wrong.

Thanks for your helpful feedback.

I put all other BOINC projects on "No new tasks", and still it took like 4 hours before any SETI@home tasks were downloaded. [4 hours after those other tasks were finished, that is.]

A bit weird, perhaps, but it's working now. I suppose I was used to the very aggressive PrimeGrid style of always sending work units immediately, vis-a-vis the SETI@home style of sending work units eventually.

And like I said, I'm very open to the possibility that this is an amateur mistake of mine, somehow.

[1 idle computer has yet to download any SETI@home tasks though. Patience, I guess.]

Part of the "issue" with PrimeGrid is that it has ridiculously short deadlines, thus they always take priority over any other project being run. Seti on the other hand has long deadlines, and so, if there any projects running at "hign priority" it doesn't get a look in for a long time.
Also, last night there were issues getting work from SETI - I think these are still on-going, this might be related to the start of the "WOW" event and people getting ready for the weekly outage later today.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

For the OP here.

Finish off all tasks in your task list, except for the last one, or perhaps two.

Updating should perhaps be enough here and not that of any reporting.

Next make sure that NNT (no new tasks) is not ticked or checked in the Projects tab of the Manager.

If nothing is happening, press Update or Refresh in the same tab, but only after at least a 5 minute delay.

If nothing is happening, wait at least 5 minutes more for each retry, or attempt.

In the end you probably will succeed, except for perhaps the scheduler not having any tasks for you and for this you should read the Event Log.

Hope that helps.

Which also means that while resource share could be set at 50/50 when running two projects, still that of the reporting deadline is what matters and next counts.

Apparently getting credit for what I am doing there, but next I am not supposed to finish a Genefer22 CUDA task the same day as it is being received.