Wrangle Cook Time

[Alain2131]

Hello !

I noticed that the Last Cook info to know how long the node has cooked for is "wrong" for Wrangle Nodes.
This can be seen easily when using the Performance Monitor to get the real data

When we look at the values in the Performance Monitor itself, we can see that the Last Cook Time for a Wrangle Node is actually only the Output VEX Code step. This seems to ignore the whole attribvop step.

This doesn't occur for other nodes, such as the blast or delete nodes.

Is this an expected behavior regarding the Wrangle Nodes ?

 

Attached is the test file, where I compare the speed of four methods to isolate the points I want.
1 - blast, cook time 0.189s
2 - delete, cook time 1.193s
3 - wrangle (pt), cook time 0.56ms ("real" is 1.632s)
4 - wrangle (detail), cook time 0.13ms ("real" is 4.189s)

At first, I thought "hey, wrangles are sooo much faster than even the blast node, so why not use them always instead"
But then, I noticed that the wait time I felt when cooking the node was a bit longer on the wrangle, despite the Cook Time telling me otherwise. This is what sparked this whole investigation

wrangleCookTime.hipnc

Thanks !

[Skybar]

Its probably because a wrangle is an HDA, and it seems to only show the cooktime for itself and not what it contains. You could always send in a bug for it!

[Alain2131]

Hey Skybar,
sorry for not updating this post, but here's an update :

I filed a RFE, and just received a mail confirming that this is a bug, and it has been logged in SESI's database; Bug (ID#95752)
She said she'll update me when the developers will have produced a fix, so I'll update this post then.

Thank you for your answer

Edited December 21, 2020 by Alain2131