Spinoff: additional statistics on projects

[ChristianVirtual]

In order to give us donors more insights on the work done would it be possible, with low effort, to provide an periodical updated list (e.g. Each 3 hours) with a summary on projects done, what hardware, what avg. TPF etc. And if possible how many WU in each projects are done or how many still to be done.

For each individual donor of course 3rd party tools can help; but over the whole project that's not possible.

[k1wi]

What about updates/reviews on specific aspects, on a less frequent (but routine) basis?

[bruce]

In order to give us donors more insights on the work done would it be possible, with low effort, to provide an periodical updated list (e.g. Each 3 hours) with a summary on projects done, what hardware, what avg. TPF etc. And if possible how many WU in each projects are done or how many still to be done.

Do you mean Projects, or WUs?

The number of WUs "still to be done" on a specific project is almost always unknown. Human judgment needs to be applied. Has the project completed enough information to draw the kind of conclusions we need to be able to draw or do we need more data? The serverstat webpage currently reports the number of WUs that are available for processing (by server, not by project) but for the most part, when a WU is completed, the number does not decrease because the completion of Gen N causes the server to generate Gen (N+1).

The number of WUs completed (by server, not by Project) is also available on serverstat, though it takes a little work to figure it out. Watch WUs Rcv over time.

TPF is not tracked. Total time from assign to upload is used to calculate the QRB but that doesn't tell if the WU was run continuously or paused for a while. Moreover, total time is not stored permanently.

[ChristianVirtual]

Statistics by server is less of interest; I would more like to know how individual performance compared to the rest of the crowd is; by project and hardware class (like GK110 or CPU:4; clock speeds might be difficult to get).
If the TPF is not permanent stored at least we could collect min, max, count and moving avg/stdev.

If there is no easy way to define a max number of WU per projects upfront it could be set to infinite; as soon some end of project is visible some estimation would be ok, too. But if there is no way, so it be.