@fgrieu although the space for FLASH is smaller than a resister, the read-out circuitry is much, much larger. It only works well because it can be shared by banks. 128-bits of FLASH is about the size of 512-bits of register (14nm SOI). That overhead is only required once though. BTW, it's probably better in older processes. 14nm SOI is weird with density as it uses DRAM for SRAM.

I really don't know what to say.

@bdegnan Yes. My reasoning applies if there is dense memory (for other reasons on the IC). BTW, is low-density memory (e.g. 1-bit "fuse") still a thing in modern processes? And can the state of such thing be read easily by optical means ?

ahh, the classic "can you do X" - "Yes"

It's about all I know about the openssl api :P

I'll freely admit that I have a tendency of answering questions exactly as posed.

ohh BTW @Maeher is there a well-defined global tiering system for crypto publication venues?

@SEJPM What does well- defined mean to you?

@Maeher a tier list that most cryptographers would produce in a mostly agreeing way

Most people I know consider the CORE ranking close enough to not bother trying to come up with something else.

@Maeher got a link to that?

@SEJPM portal.core.edu.au/conf-ranks

@Maeher ty

@fgrieu The "e-fuse" is alive and well. You use these for IC serial numbers, and things were you wouldn't want it erased. You can optically see when a fuse is blown. Intel went away from them, but the chinese were decapping, and then using UV to erase. You generally remove the erasing circuits so they are "write once". Anyway, they quickly went back to e-fuses.

@bdegnan Thanks for that update !

@fgrieu e-fuse blown: semiengineering.com/wp-content/uploads/2016/12/kilopass1.png

The nice thing about floating gates is that you cannot "see" the state optically, and if you put a probe to them, you lose the charge. You have to check the currents created by the floating-charge through the device.