Maybe an extended April Fools, but this OET percent progress commenting is ongoing in 2 threads. Maybe return this thread to what the title says... the [True] milestones thread

2 Gold Sovereigns

86% Jun 2016

Greetings,

Our estimates are based off the average number of batches loaded into BOINC for the past 30 days. Currently we are averaging 46 batches loaded per day. We also take the total number of batches BOINC estimated to run for the project, currently 5463 for OET1. Then to determine how many days are left, we subtract total number of batches Loaded and Completed from that estimate and divide by average batches loaded per day. As of the other day that was 59 days left of work.

The rate at which batches loads is based on how fast things are being run on the grid as well as how many work units are inside a given batch. Before batch 1929, we were getting batches with 100k work units per batch. These work units were of flexible type which take longer to run. The newer batches are of ridged type and run faster. There are currently only 1k-2k work units packaged in a batch. From these two changes taking place, you can see that the estimate will speed up quickly.

We always reach out and discuss with the researchers how many estimated batches they expect. Currently we do not have an estimated updated number of batches for OET1, thus the estimate appears to be correct of an end date around June 4th.

If the estimated number of batches increases or we get more ridged work units before the end, then this estimate will change again.

Thanks,
-Uplinger

Keith, thanks for the explanation.

CJSL

Crunching for a brighter future....

Greetings again,

I have received more information about the batches coming down the pipeline. We are still estimated to run the 5463 batches, no update to that number. But we are about to hit more of the flexible work units that everyone was used to earlier in the year. This means that batches will begin to have the 100k work units. This will of course slow down and push the estimate back. We also have 200 batches that will be run towards the end only on PC as they will be monsters. From that information alone, my personal estimate is that we will have work for the rest of the year and into 2017.

Also, since the % complete is based off of time instead of actual batches, when the estimate starts getting longer out in the future, the percentage complete will decrease. So please do not be alarmed when that happens.

If you want to know why we used time instead of batch counts for % complete, that is due to the amount of data in the database that gets archived/purged over time. The start date of the project as well as estimated end date are two fields in a single table.

Thanks,
-Uplinger