I'm having some conflicting thoughts on how I'm interpreting the charts of two Streamline captures of the same scenario over 4s (from a Mali G-71; Samsung A20e) , so I was wondering what's your view on the following?
For instance, while the number of L2 and external texture reads per cycle have increased (from Before to After), the total amount of texture bytes read from both the L2 cache and external memory is lower.
My interpretation is that in After we're performing less filtering operations but those are more bandwidth expensive (less coherent or heavier texture format?) thus the increase in bytes/cycle. However, all-in-all there's still an improvement because in total we're reading less data. Is this a reasonable read of these?
On the flipside, there's L/S reading. There's an overall (positive) drop in all the metrics (and increase in the full read cycles). So, the L/S reading seems clear to me, compared to the texture unit one.
Any views are appreciated. Thanks!
FYI: Edited in an additional comment on the LS counters above after you hit accept.
Cheers! It actually refreshed while I was finishing reading - I was just happy with the answer at that point, but it certainly made it more complete :).