state of PIC SRAM on boot

This is totally wrong.

These is _no_ RAM of any kind that will maintain its values on power down.

ROM yes, RAM no.

So, you did not read this anywhere.

Some RAM chips may power up in the same state as the last time it was powered up, but it is not guaranteed.

donald

This is totally wrong. FRAM, and core - for similar reasons - maintain their values on power down.

It's not guaranteed to be anything in particular. Chances are it's far from random, and if the power down was brief, or incomplete, probably influenced to a certain extent by the previous states.

Best regards, Spehro Pefhany

FRAM != RAM

Guess what:

"FRAM offers features consistent with a RAM technology, but is nonvolatile like a ROM technology. FRAM bridges the gap between the two categories and creates something completely new -- a nonvolatile RAM."

Parallel FRAM looks pretty much exactly like SRAM from the system's perspective; it is a drop-in replacement. e.g. is a direct drop-in for a 62256 32Kx8 SRAM.

Not that the OP isn't on crack, of course, but your statement isn't absolutely right, which is why it irked me to see it advertised as such.

and to confuse you all even more, have a look a Maxims high security "Low imprinting" SRAM technology.

Yes, seems SRAM is not truly random on power off, but does have a statistical bias based on what was there last time.... Not good enough to get 100% data back, but I'm sure cracking teams would love to have a 'better seed' than a random starting point.

Not likely to matter much, in the average PIC project :)

-jg

Thanks for straightening me out everyone. I was hoping it would be either maintained or in a somewhat random state so I could use it as a seed to a PRNG.

If you need to store something many PICs have built in EEPROM or Self programable Flash

There are ways to take semi-random data and create random data from it using some kind of hash. Proably the simplest is applying an XOR function something along the lines of:

int gen_seed(int array[], unsigned length) { int value = array[0];

while(--length) value ^= array[length];

return value; }

Provided that length is large enough, any underlying tendencies of memory bits to favour one value or the other are soon swamped by the randomness present in other bits. I have a mathematical analysis of the above code somewhere that states how long length must be to create a truly random number, based on how strong the tendencies are. I'll have a look around and see if I can dig it up.

Of course, hardware random number generation needn't be particularly complex: see