Individual overall statistics graphs

[Chuck]

Has something happened to the individual overall statistics graphs to reduce the GPU TF amounts for the past several days? Mine seemed low, and when I looked at several other users' graphs they also seemed low.

[kladner]

I noticed this also, but I think the relative scaling between CPU and GPU shifted a bit. I don't have any captures of the previous state to check this, but I'm sure chalsall will fill us in.

[bcp19]

Originally Posted by kladner

I noticed this also, but I think the relative scaling between CPU and GPU shifted a bit. I don't have any captures of the previous state to check this, but I'm sure chalsall will fill us in.

I could swear that mine was averaging the last few days worth instead of tracking the actual days performance (I had one machine down for 2 days)

[chalsall]

Originally Posted by Chuck

Has something happened to the individual overall statistics graphs to reduce the GPU TF amounts for the past several days? Mine seemed low, and when I looked at several other users' graphs they also seemed low.

Nothing has changed there (I don't think). Keep in mind that for several months now returned results' dates are "adjusted" to spread them out across the period of time they were actually checked out, in order to make each day's efforts be approximately the same. This was to prevent the extreme spikes which were caused when people would check out large number of assignments, work on them for a couple of weeks, and then report them all at once.

But it's always possible I broke something in the back-end sub-systems which does all this work. Let me look into it a bit and report back.

[Chuck]

Originally Posted by chalsall

Nothing has changed there (I don't think). Keep in mind that for several months now returned results' dates are "adjusted" to spread them out across the period of time they were actually checked out, in order to make each day's efforts be approximately the same. This was to prevent the extreme spikes which were caused when people would check out large number of assignments, work on them for a couple of weeks, and then report them all at once.

But it's always possible I broke something in the back-end sub-systems which does all this work. Let me look into it a bit and report back.

I think what I saw must have been the results of the spreading of the data.