Is flash drive zeroed or translator wiped ..

[Joep]

I had this idea we might be able to detect somehow without chip-off to see reading back zeros or FF pattern may be due to a pattern written or wiped translator.

Hypothesis being: I expect wiped translator to result in faster read-speed than reading back a pattern that was written and possibly power consumption of reading from wiped translator < consumption of reading an actually written pattern. If we look at voltage needed for various flash operation we see order of read, write, erase. But even for read a charge pump needs to up voltage by a bit and so it may give measurable increase in power consumption. This is my somewhat layman understanding, feel free to correct me (@fzabkar). If we consider wiped FTL the controller may not even need read the NAND.

In advance I need to say I need to repeat this test with a faster UFD. This is an ancient UFD, USB 2, so USB bandwidth may be troubling results of read back speeds.

I don't see difference in read back speeds, though the flatter line in case of wiped FTL may be worth investigating:

zero-fill.png (34.51 KiB) Viewed 5769 times

ftl-wipe.png (34.39 KiB) Viewed 5769 times

Difference in power consumption is small but measurable ..

Reading back from zeroed drive:



Reading back from drive with FTL wiped:




I will follow up when I have more tests with USB 3.x UFD. Perhaps I should get a measure apparatus that can save data so I can plot stuff.

[Joep]

Perhaps also mildly interesting difference in consumption between writing zeros and wiping translator .. This also sort of led me to conclusion the MPtool is wiping translator probably because consumption NAND erase > consumption writing zeros.

Zero filling:




Wiping translator (if I disable option to verify the NAND, consumption is lower even):

[Joep]

I got one now that logs. But still the wrong one found a higher resolution one now. So will go buy that one too.

zeroedvstrimmed2.png (26.32 KiB) Viewed 5393 times

[pclab]

Nice investigation.