Project 18706 - Timeout too short

[Gary480six]

Can someone check on the benchmark for Project 18706. I think the timeout was set too short when the project came online.
It has 188,026 Atoms but the Timeout is only one day.

Project Atoms Timeout

18706 188,026 1 day

18477 90,548 1.5 days

19012 90,246 1 day


As you can see, projects with half as many atoms normally get a 1 to 1.5 day Timeout... or more.

What I'm finding is that because the Timeout was set too short, some of the older video cards are missing the Timeout deadline. That means the same work will be sent out again to be completed by another volunteer.
That's a waste of our resources.

[toTOW]

Timeout value and the number of atoms is completely unrelated. There's another parameter to consider (and that you can't see on psummary) : the simulated time in each WU.

If project A with X atoms ran for 1 day and project B with 2*X atoms would run twice slower, then its simulation time could be set to half the one set in project A, and it would still run for 1 day.

[BobWilliams757]

Just an FYI on this project, with no regards to how the timeout or deadline are set.

I picked one up on my lowly Vega 11 onboard my Ryzen. It could make the deadline.... barely. I assume that due to the AMD cards not having species identification that my onboard graphics are lumped in with quicker Vega stuff, so I just roll with it.

In this case I ran it, by the time I realized the timeout had passed it was only a few hours from the deadline and appeared it would make it. So I let it fold thinking maybe it would still be the first done. Someone else had picked it up after the timeout and finished it a few hours before mine. Oh well, at least I tried.

[BobWilliams757]

Another update, my previous comment used poor language.

Not even close to meeting the timeout, and barely meeting the expiration on my system.

Looking at the LAR website, it appears that the species selections for 18706-07 may be optimistic for both AMD and Nvidia GPU's.