Maximum writes to M2 EEPROM

rq3

Senior Member
These are always fun experiments... Keep us posted. Thanks.

Also, how long does your code take to do the read-and-verify loop? How did you calculate the 10 billion reads?
It's a total SWAG on my part, but I figure I'm within an order of magnitude at 10000 reads per second (32 MHz clock). Roughly 6 billion reads per week.
 

hippy

Technical Support
Staff member
That's a lot of reads but was the strategy sound ?

My understanding of the 'read disturb' problem is that reading one cell degrades its neighbours but also restores the one read to full health, assuming it hasn't degraded too far. Thus you need to do a lot of reads, but not from the cell you are looking to destroy. You can only read that once it should have been destroyed.
 

rq3

Senior Member
That's a lot of reads but was the strategy sound ?

My understanding of the 'read disturb' problem is that reading one cell degrades its neighbours but also restores the one read to full health, assuming it hasn't degraded too far. Thus you need to do a lot of reads, but not from the cell you are looking to destroy. You can only read that once it should have been destroyed.
At start, I'm basically writing zeros into 6 bytes (3 words), and ones into 6 different bytes (3 words). Then I repetitively read all six words. If any of them change, the LEDs go off and the program quits. Of course I have no clue whether the physical silicon is laid out anything like adjacent bytes per cell, so this is likely a huge waste of time. But interesting (to me anyway).

Rip
 

rq3

Senior Member
That's a lot of reads but was the strategy sound ?

My understanding of the 'read disturb' problem is that reading one cell degrades its neighbours but also restores the one read to full health, assuming it hasn't degraded too far. Thus you need to do a lot of reads, but not from the cell you are looking to destroy. You can only read that once it should have been destroyed.
Given today's NASA report of NVRAM issues with their 2004 Mars rover, I've decided to house my test circuit in a 0.1 Torr carbon dioxide environment and launch it to an altitude of 62 miles with a mylar balloon for a period of 10 years. That should get some results!

Rip
 

Dippy

Moderator
Yes, I think I said that last year.
(It always sounds more dramatic when the calender rolls over :) )

And just think how many fetch reads you're doing with the programme running... does that count?
All good stuff.
 
Top