[Cryptech Tech] RSA blinding
bernd at net2o.de
Tue Jul 4 22:31:26 UTC 2017
Am Mittwoch, 5. Juli 2017, 00:23:40 CEST schrieb Pavel Shatov:
> In step 6. block memory R is either overwritten with its own contents
> (that changes no bits) or with a different value that on average changes
> half of the bits. I have a feeling that those will have different power
If you want to avoid that, add one bit per memory cell which means "xor the
rest with a pattern that has half of the bits set ('h55... 'haa..., 'h33,
'hcc, all are equally good). That way, you can achieve true constant-power
for the memory write at the cost of just one bit per memory word.
However, most of the energy going into SRAM writes is precharging the bit
lines, and discharging them, and since those are all symmetric (you discharge
either the left or the right bit line for a SRAM write), only the flip/non-flip
of the SRAM cell itself can make a difference. And that difference is small
compared to the rest.
"If you want it done right, you have to do it yourself"
net2o ID: kQusJzA;7*?t=uy at X}1GWr!+0qqp_Cn176t4(dQ*
More information about the Tech