CacheAsBitmap: screen-res ( meaning post container transforms) or not?

Question

Here is a question I'd love to find the answer to:
1.  Suppose you have a Sprite which contains a shape instance within.
2.  The sprite is scaled down ( let's say 50% ), and added to the stage, but the shape it contains is left unscaled.
3.  Now, if the shape has CAB enabled, is the cached bitmap rendered at the final screen res, post container transforms ( in this case, the cached bitmap would be 50% smaller than the shape ), OR Is the cached bitmap rendered at 100% the size of the shape, and then downsampled when it is drawn onto the screen?
...why is this relevant?
For example:
A)  IF it turns out that cached bitmaps resulting from CAB are screen-res ( post all container transforms ), then it might make sense to set CAB to true even on a Bitmap instance, if one of its parent containers is scaled up or down ( so as to minimize the resampling effort, every time the bitmap is drawn to the screen ).
B)  In the earlier example, with the unscaled shape within a scaled sprite, IF it turns out that cached bitmaps resulting from CAB are NOT screen-res ( meaning NOT post container transforms), then it might be advantageous to move the shape out of its container and apply the 50% scaling on it directly ( again, to minimize the resampling effort required to draw the cached bitmap onto the screen ).

Answer 1

Here is a question I'd love to find the answer to:
1.  Suppose you have a Sprite which contains a shape instance within.
2.  The sprite is scaled down ( let's say 50% ), and added to the stage, but the shape it contains is left unscaled.
3.  Now, if the shape has CAB enabled, is the cached bitmap rendered at the final screen res, post container transforms ( in this case, the cached bitmap would be 50% smaller than the shape ), OR Is the cached bitmap rendered at 100% the size of the shape, and then downsampled when it is drawn onto the screen?
...why is this relevant?
For example:
A)  IF it turns out that cached bitmaps resulting from CAB are screen-res ( post all container transforms ), then it might make sense to set CAB to true even on a Bitmap instance, if one of its parent containers is scaled up or down ( so as to minimize the resampling effort, every time the bitmap is drawn to the screen ).
B)  In the earlier example, with the unscaled shape within a scaled sprite, IF it turns out that cached bitmaps resulting from CAB are NOT screen-res ( meaning NOT post container transforms), then it might be advantageous to move the shape out of its container and apply the 50% scaling on it directly ( again, to minimize the resampling effort required to draw the cached bitmap onto the screen ).