If you look near the bottom of the menu on the left side of this page, you'll see a new link, "Task status".

There's charts on that page showing the number of outstanding tasks in each subproject, and the recent execution times for each subproject.

Most of the charts use a logarithmic scale on the y-axis due to the large differences between different subprojects.

Please note that execution times are expressed in hours. Times less than one hour are shown in thousandths of an hour, so "500m" is half an hour, or 30 minutes.

You may thank JimB for the idea, and blame me for the execution.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

Great addition!

Looks good :) Will be more interesting as data builds up for longer term trends.

I have to ask, what does "task in progress" mean? Does it mean "has been sent out but not yet returned"? As opposed to, new tasks sent out, or tasks received?

For example, say I do short units (SGS, PPSE) without cache, and return them as soon as they're finished. Would that show up as one "outstanding" unit per core, even though I might do hundreds per day?

Reason for asking is I think it is a more interesting number to see how many units are "done" than how many are floating out there which can vary due to cache and unit lengths.

I have to ask, what does "task in progress" mean? Does it mean "has been sent out but not yet returned"? As opposed to, new tasks sent out, or tasks received?

For example, say I do short units (SGS, PPSE) without cache, and return them as soon as they're finished. Would that show up as one "outstanding" unit per core, even though I might do hundreds per day?

Reason for asking is I think it is a more interesting number to see how many units are "done" than how many are floating out there which can vary due to cache and unit lengths.

These statistics are gathered at 5 minute intervals, and the database query needed to gather the data for "how many do we do each day" is (currently) quite expensive. I may add this if I'm able to optimize the SQL to an acceptable level.

These statistics are gathered at 5 minute intervals, and the database query needed to gather the data for "how many do we do each day" is (currently) quite expensive. I may add this if I'm able to optimize the SQL to an acceptable level.

To further complicate matters, I was thinking how would you count completed tasks anyway? Returned tasks may include error, aborts and timeouts. Other tasks may look ok but later invalidate. Waiting to validate would delay the stats. I was thinking it may be adequate just to count correctly formatted results, like the stats for challenges, even if they may later be invalidated. I'm assuming the invalidated units would be relatively low in number so not significant as far as these stats are concerned.

I have added a "New Users" chart that shows the number of new users within the last 24 hours. It shows two metrics: all users, and "active users". Active users are users that have started to run at least one task.

I suspect that non-active users are mostly spambots, but I have no data to support that assumption.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

I recently added a T-Flops/s chart to the new page. It shows the current processing power in use by all of you in support of PrimeGrid. It's currently about 1.2 Peta-Flops per second.

I wouldn't try comparing our Flops estimate to anyone else's. Chances are any other BOINC site calculates them differently than we do, and who knows how non-BOINC sites calculate this number. The purpose of the chart is mostly to see how our processing power changes over time, especially during challenges as well as long term seasonal and yearly trends.

I can, however, tell you how I'm doing the calculation:

200 credits is, by definition, equivalent to a day's worth of processing by a computer doing 1 Giga-Flop per second. Our Flops estimate is therefore based on the number of credits granted. I think we have one of the most accurate credit systems, so it stands to reason that the Flops estimate should be pretty accurate also.

Note that this number includes only BOINC tasks. It does not include the processing power involved in either manual sieving or PRPNet.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

200 credits is, by definition, equivalent to a day's worth of processing by a computer doing 1 Giga-Flop per second.

What makes me wonder -- is The Long Job Credit Bonus considered in or out of these 200 cobblestones?

I think this is a great addition. I've always been curious as to which projects get more use than others. I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.
____________

I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.

I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.

Those charts are already there. :)

I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.

Those charts are already there. :)

I missed your post saying it had been done, and didn't scroll far enough down the page to find it. Very cool! I'm moderately surprised that there are more ESP sieve tasks finished per day than PPS sieve tasks. It's very interesting to see how these trends evolve over time.

I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.

Those charts are already there. :)

I missed your post saying it had been done, and didn't scroll far enough down the page to find it. Very cool! I'm moderately surprised that there are more ESP sieve tasks finished per day than PPS sieve tasks. It's very interesting to see how these trends evolve over time.

I would like to also see the statistics on number of tasks completed per day, but do understand how that would be more intensive to query from the database.

Those charts are already there. :)

I missed your post saying it had been done, and didn't scroll far enough down the page to find it. Very cool! I'm moderately surprised that there are more ESP sieve tasks finished per day than PPS sieve tasks. It's very interesting to see how these trends evolve over time.

I actually find it pretty normal. PPS runs one pre GPU, so usually 1 or 0 (I can imagine a plethora of older PCs falling into this category). ESP, on the other hand, runs 1 per core, so usually 2 or more (and it's the most advised project for that plethora of old PCs, as well as AMD's). Besides, now that GFN has some pretty quick tasks, I'd expect GPUs to focus on those instead.

Quick question, on the completed tasks charts is that tasks that have been turned in and may or may not be pending validation or is it just tasks that have been turned in and validated?

Cheers Rick