No new work is being ignored by gpugrid as well as seti and seti beta. I just discovered this when I tried to run down my gpugrid backlog so I can start something else. I have known since 6.6.1 that the NNW was being ignored and assumed it was BOINC. Now gpugrid is also doing it and I asked about it on their forum.

Anyway, is ignoring the NNW a BOINC "bug" or a project "feature".

BM 6.6.4, Vista64

...I have known since 6.6.1 that the NNW was being ignored and assumed it was BOINC. Now gpugrid is also doing it...

??????????????
Of course it is BOINC. Who else???? All the BOINC projects use BOINC software, and that was screwed up for CUDA.

Gruß,
Gundolf
[edit] And BTW, 6.6 is alpha software, so you have to expect such errors.[/edit]

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

Of course it's BOINC. But I think it would be helpful if we could work out whether it's the BOINC client (which the user can do something about - enable logging flags and report behaviour, in detail, to the developers: or just downgrade to a stable version), or the BOINC server (where all the user can do is send advice to project admins to update their server software, if we can identify which changes the admins need to apply).

FWIW, I think I've seen both client problems and server problems reported on the boards. I even got caught myself by a server problem at SETI (Work fetch anomaly). In that case, we were told that the server issue had been fixed: we looked for, but couldn't find, any relevant changes in the server code base: but the problem seems to have gone away. It's difficult to debug "Server sends too much work" problems at SETI, because often enough they can't generate enough work to fulfil legitimate requests - but work consumption has entered a quieter phase recently, so we might be able to find out whether this particular problem has been permanently laid to rest.

This is a back end problem, people. It's the scheduler on the server that is ignoring your zero work request. When you see things like this happening, report it to the project.

---

This is a back end problem, people. It's the scheduler on the server that is ignoring your zero work request. When you see things like this happening, report it to the project.

Why should the server people fix code that is working just fine for them? They send a WU that I dont want, it gets processed at high priority, and the result get sent back quickly.

This is a variation of the "Oil Cartel Dilemma" where one project cheats and sends out work that goes for a high price while the other projects honor the agreement and start losing in the battle for the big gpu's.

my 0.02c ;-)

NNT is working correctly with my Win2003, BM 6.6.5, non-service. Recent part of my log is posted here

...my 0.02c ;-)

That's really a pretty little amount! (0.0002$ ;-)

Why should the server people fix code that is working just fine for them?

When the server is sending you tasks while you aren't requesting any, the code isn't working correctly. Your BOINC tells the scheduler that it is not asking for any work, the scheduler answers with "that's fine, but here, have some anyway". But you want the BOINC client to be fixed? What should be fixed then? Go explain that to me.

---