Essential difference between normal irda mode and low-power irda mode?

guys, I'm learning stm32f401x mcu. All I have is an rm0368 reference manual pdf file. No mcu chip right now. From the manual, I learnt that PSC instead of USART_BRR is used to generate baud rate in low-power irda mode. I'v also read the irda spec pdf file, but it didn't help (perhaps I just missed important things). So what's the essential difference between normal irda mode and low-power irda mode, from the point of view of usart circuitry? Does the term 'low-power' refer to the entire irda module in a finished product (due to weaker radiation power), or does it refer to the irda block inside the mcu chip (due to shutdown of certain sub-block under this mode) ? Thx.

---- a related question on pulse generation in low-power irda mode -------

In normal irda mode, we use USART_BRR to generate baud rate from PCLK. So the pulse width is basically 3 * (Tpclk * DIV_Mantissa), DIV_Mantissa is a 12-bit long number.

While in low-power irda mode, PSC[7:0] is used to generate irda low-power baud rate from PCLK. The manual says "Instead, the width of the pulse is 3 times the low-power baud rate which can be a minimum of 1.42MHz". I assume that in this mode, the pulse width we can obtain is 3 * (Tpclk * PSC), where PSC is an 8-bit long number.

How can I get a longer pulse with a smaller divisor?