I want to your advice about i.MX6Q SDMA.
Our partner is using i.MX6Q SDMA for ESAI on a custom board, but its performance is not enough currently.
In fact, they want to over 54 FIFOs read access from ESAI when Fs=48kHz, but now SDMA seems to not be able to handle over 54 FIFOs read access.
So we want to information how to improve SDMA performance.
Please see their environment and our questions as below.
DRAM: DDR3 528MHz 64-bit
CCM: Default as Linxu BSP (L3.0.35_4.1.0)
They are using app_2_mcu for ESAI now, but shp_2_mcu seems faster than app_2_mcu.
Should they use shp_2_mcu instead of app_2_mcu to improve SDMA performance?
If the answer to Q1 is "YES", would you let me know whether Linux BSP (L3.0.35_4.1.0) supports shp_2_mcu for ESAI?
Could you give me advice to improve SDMA transfer performance other than using shp_2_mcu?
(e.g. They should increase the number of Buffer Description, they should write the data from ESAI to on on chip SRAM instead of DRAM, etc...)