SPRUIV7B May 2022 – September 2023 AM620-Q1 , AM623 , AM625 , AM625-Q1 , AM625SIP
Access time to a NAND memory device can be optimized for back-to-back accesses if the associated nCS signal is not deasserted between accesses. The GPMC access engine can track prefetch engine accesses to optimize the access timing parameter programmed for the allocated chip-select, if no accesses to other chip-selects (that is, interleaved accesses) occur. Similarly, the access engine also eliminates CYCLE2CYCLEDELAY even if CYCLE2CYCLESAMECSEN is set. This capability is limited to the prefetch and write-posting engine accesses, and accesses to a NAND memory device (through the defined chip-select memory region or through the GPMC_NAND_DATA_i location, where i = 0 to 3) are never optimized.
The GPMC_PREFETCH_CONFIG1[27] ENABLEOPTIMIZEDACCESS bit must be set to enable optimized accesses. To optimize access time, the GPMC_PREFETCH_CONFIG1[30-28] CYCLEOPTIMIZATION bit field defines the number of GPMC_FCLK cycles to be suppressed from the following timing parameters:
Figure 12-206 shows that in the case of back-to-back accesses to the NAND flash through the prefetch engine, CYCLE2CYCLESAMECSEN is forced to 0 when using optimized accesses. The first access uses the regular timing settings for this chip-select. All accesses after this one use settings reduced by x clock cycles, x being defined by the GPMC_PREFETCH_CONFIG1[30-28] CYCLEOPTIMIZATION bit field.