Cloth computation conundrum

[robcat2075]

Here is a screen capture of the status bars of two instances of A:M

They are both running the same cloth simulation with the same settings.

The lower left one highlighted in green was started first. The upper right, in yellow, was started second and quite a bit later.

The first instance (green) is bogged down in absurdly high numbers of substeps for each frame. The second instance (yellow) is sailing through with far smaller substeps needed for each frame and getting done faster.

Same computer, same project, same settings. 

 

[robcat2075]

 

Less than an hour later the first one has barely advanced. The second one is far ahead.

[robcat2075]

However... a few hours later the second one had exploded in thousands and thousands of collisions and was never going get past 77%

The first one was still chugging along slowly completing frames and had made it to 33%

Unfortunately, stopping one stops both.