In other words:

The start of the new UGM1 project is near.
Keith, we really don't tell it to others ears.
The secret will stay in BETA forum here.
I don't see the forest through the trees.

I am currently running 15 of the work units that errored out due to high disk usage on our alpha grid. I am seeing a few results that are 20% complete with 200MB result files uncompressed after 4 hours. Our goal is to not have this happen in the future.

Thanks,
-Uplinger

Just thought I would throw in my two cents worth from Uplinger's oh so helpful figures. 96.29% of the Beta WU's finishd in 10 hours or less. 90.61% of the total Kbytes generated were from WU's which finished in 10 hours or less with a full 50% of the total generated by those which finished in 4 hours or less. I could not find any pattern in "average kbytes generated per hour." It would be interesting to see a breakdown of time for various cpu's. From this very preliminary data it looks like a majority of the potential WU's will be less than 5 hours.
Lets see what the next round of beta brings.
Cheers

Sgt Joe,

The report with sizes and avg sizes is a little bit left over from previous database lookups I have done. Those are more related and hold more value when you compare against batches. The reason for that is we used different datasets for each group of work units. Thus one dataset creates results larger than others. This is not technically a bad thing as it is within reason for the workunits. However we are trying to determine a way to prevent such large swings even within a batch. For example we had some results within that batch that were less than 100KB (compressed) result file, but a handful of others grew above 250MB (uncompressed). I am currently running a work unit that has created 375 MB (uncompressed) and it is only 37% complete. In theory the max a workunit sized this way could be is 31GB (uncompressed), but we should not see data that large.

Yarensc,

We have people who sign up all sorts of machines. On average it is slightly faster than the grid, but in general it is a wide range. We have members who attempted to run Pentium 3 processors against this latest beta. In which the processes failed due to this project requiring SSE2.

Thanks for the feedback and anhhai suggested, we are thinking about running another round of BETA with different workunits. But again, no ETA on that.

Thanks,
-Uplinger

I see 6 WU's on my machines.

5 validated
1 pending validation, waiting on wingman

run times varying between 3:12 and 7:44.