STM32U5A5 Semp put/get speed test

[Riscy]

I have callback ISR code (from UART) where I put in Semaphore put and other code with Threadx static implementation with stack (SRAM2), Semaphore get, etc. I placed two GPIO signals to measure the timing between put and get. I set the compiler to optimize for speed setting and used Logic Pro 8 to probe these two GPIO. The thread priority setting is 1.

I have other 3-4 treads running flashing LED but at a lower priority (around 10). 

I found that STM32U5 while running on 160MHz to be 2.6uSec between semp put and semp get. 

Microsoft spec claims 0.2uSec for 200MHz of something MCU.  Have I missed something?