Multiple reports of the Spi_AsyncTransmit() function taking longer than expected.

From DesaySV:

Do you have any comments about why the “Spi_AsyncTransmit” code execution be slower when DDR throughput higher?

After change the code to MSRAM from DDR, the efficiency only improved 30%, is it reasonable?

From Reachauto:

And the time consuming part is Spi startSeqAsync,It check all the hardware queue and consume any pending job if the hardware is free.

As i test,when i use one HardWare,it costs 40us .I'd like to know whether its time can be reduced by any operation, or it can not be optimized?

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1088080/tda4vm-the-spi-transmission-takes-too-long-used-qnx-with-pdk_jacinto_08_00_00_37