From this, you have to be at point blank range with a rapid fire autocannon to be able to kill a T-72/T-90 with it.
(The T-90 is actually a rebadged T-72 with a new turret in later marks).
Essentially, as DRG found out:
the armour of the side of the hull is 80mm thick. The armour on the sides of the engine compartment is 70mm thick.
War Thunder (I know, I know); lists the 30mm 2A42 Autocannon as having at 100m the following penetration:
HE/AP: 63~mm
APDS: 81~mm
The penetration of HE/AP is deep enough that there would be scabbing off the inside of the armor; especially in the rear 70mm thick armor around the engine.
So you can see that in order for a BMP-3/BTR-4 to stand a chance of defeating a T-72/90; the following must occur:
1.) It must be within 100m or less.
2.) The enemy T-72/90 must not have side skirt armor (or be missing pieces of it).
3.) The gunner has to recognize the missing side armor as a weak spot.
4.) The gunner then has to place 20~ rounds into that area.
That's a lot of conditionals that have to occur for a "one in a million" shot.
People need to remember the scope of Steel Panthers -- it's not a "down in the weeds semi accurate penetration simulator" like War Thunder is; but more of an abstraction of armored fighting; i.e. the game already abstractly accounts for weak spots; just load
MBT_APCalc and look at Russian Weapon 77 - 30mm 2A42 A/C:
100m Range:
AP: 6 typical penetration, 7 best possible penetration
Sabot 8 typical penetration, 11 best possible penetration