[BUG-4182] [Improvement] The new cap to Avatar Render Weight needs to be done post trend calculation to accurately reflect data as well as large changes. #13177
Comments
Lucia Nightfire commented at 2013-10-16T22:25:55Z, updated at 2013-10-17T06:25:45Z My 20 minute estimate was way off as it would take probably 1.5 - 2 hours to get near 500k. The following two passes were done with a target ARW >1.2 million, both values different from each other. [13:03:23]: Target render weight is now 97910. Was -1 [13:57:22]: Target render weight is now 97910. Was -1 The importance of the data is in the change in trend. Which is why not capping the value entering the trend is important as currently, anything greater than 500k is trended the same way as these passes show since the data mirrors one another despite two different ARW values. Currently a value of 500k and 20 million will trend the same. Agents entering a region start the trend at whatever ARW they have (capped at 500k). |
Simon Linden commented at 2013-10-17T17:58:20Z FWIW the rate of change will also depend on the number of viewers making reports to the server. |
Information
As of server version 13.10.15.282406, ARW values sent by viewers are now capped at 500k before that value is factored into the trend data.
This pre-capping now causes an inaccurate trend.
This is now having an even more undesired affect of reflecting when >500k changes to ARW occur.
Example:
Someone thats been in a region a few minutes that has a ARW of 50k attaches a sculpt lagger that causes their ARW to jump to 3 million.
Pre-Trend Cap:
With the current pre-trend capped value, it would take over 20 minutes for their reported ARW to get near 500k.
Post-Trend Cap:
With the preferred post-trend capped value, their next reported ARW would be 500k because the 3 million would be the value factored into the trend data.
Please make the cap, post trend instead of before. Please allow the trend to accept values up to max integer/float. Thanks.
Links
Related
Original Jira Fields
The text was updated successfully, but these errors were encountered: