2
\$\begingroup\$

We're working on a new project, and we're considering using a TMS320C6657 DSP, or a similar DSP. There are many boot modes, NAND Flash, NOR Flash, Ethernet, etc.

What are the pros and cons of booting from NAND vs NOR ? As far as I know, NOR Flash memories can't have bad blocks so we're tilting towards NOR Flash.

\$\endgroup\$

1 Answer 1

2
\$\begingroup\$

A common configuration is a smallish serial NOR (SPI/QSPI) for the first stage boot, followed by a NAND device on a mass storage interface such as ONFI, eMMC, SD card, UFS, USB, etc. for the kernel and application code.

With the stuff divided up this way, the NOR holds the needed support for NAND, including the IO driver, bad block / ECC management and so forth, in a medium that doesn’t require error management itself. A major benefit of this split approach is if the NAND gets corrupted, you always have a means to un-brick the system with what's in the NOR.

That said, there are NAND devices on offer in QSPI or octal SPI with built-in ECC to address the error recovery required for robust boot, while providing enough room for kernel and application code. More here: https://www.jblopen.com/qspi-nand-intro/

A related NAND interface for embedded is Hyperbus, which is based on octal SPI but a bit faster (and it's a JEDEC standard.) Xccela is Micron's version of breathed-on octal SPI. These would be good choices for a newer system.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.