why devide by 32 while calcluting baudrate in 8051

The important thing is that a UART needs a clock frequency that is several times faster than the baudrate.

This allows the UART to support a mismatch in baudrate between sender and receiver.

The UART samples again and again until it detects the start of the start bit. Then it keeps counting n clocks / bit until having taken care of the stop bits.

While counting the clocks, the UART takes a couple of samples within each bit window and uses a majority vote to determine if the bit was a zero or a one, improving noise handling.

And by ignoring the start and end of each bit window, the UART can handle that the bit window slowly glides forward or backward because of smaller baudrate errors.

If looking generally at UARTs, it is common that the UART samples 16 times faster than the baudrate. Some may sample faster and a few lower-quality UART may sample slower.

Thank you