SPRUIM2H May 2020 – October 2023 AM2431 , AM2432 , AM2434 , AM6411 , AM6412 , AM6421 , AM6422 , AM6441 , AM6442
This mode is used for Direct Read Mode of operation. If any other operations are intended to be executed, it is recommended to disable PHY Pipeline Mode and re-enable for subsequent Direct Reads in PHY mode. Since there is comprehensive software mechanism controlling Read data transfers in Indirect Mode, pipeline of data interface accesses is not effective for this mode. Enable PHY Pipeline feature when at least four 4-byte-sized data words are predicted to be read in sequentially. The Flash Command Generator pipelines and puts them into TX FIFO which causes CS to remain active because low level SPI protocol controller controls TX FIFO fill level. In order to correctly trigger Direct Read in Pipeline Mode TX FIFO must be empty. Therefore first polling of OSPI_CONFIG_REG[31] IDLE_FLD bit needs to be done. The sequential data transfer will be interrupted when the data target select signal of the data interface is asserted to low. This information is also detected by Data Target Module which informs the Flash Command Generator that the next access is invalid and a TX FIFO locations can be flushed transparently for the system.
In PHY Pipeline Mode it is recommended for Data Controller not to introduce wait states in between consecutive occurrences of the data interface signal that indicates transfer has finished. It will ensure regular transfer rate on data side. Introducing wait states gradually slows data transfer rate down and may finally cause SPI transfer interruption because of TX FIFO data starvation. The system, however, may need to introduce some number of wait states after completion of sequential transfer (composed of 4-byte sized data words) for progressing the data. The dedicated buffer is implemented in Data Target Controller which collects all incoming data during wait states injection. In order to keep SPI transfer uninterrupted, number of wait states should be as little as possible. The higher the OSPI_HCLK/ OSPI_RCLK ratio the more wait states can be introduced without SPI transfer interruption. In case the system is able to launch a new transfer before wait states overflow, buffered data transfer to the host will continue. It compensates slowed down transfer by introducing wait states. In case the system is not able to launch a new transfer before wait states overflow, next incoming transfer is considered non-sequential and is executed after all pipelined data is flushed.
This mode can be enabled when following conditions are met: