Re: [Suit] Fwd: Firmware Update Paper

>
> The time we measured spent on hashing (SHA256)is comparatively small for
> the typical firmware sizes.It takes ~4ms per kB to hashon the platform
> used in the paper. Even with a 200kB binary,it takes just a fraction
> (~10%) of the time needed for signature verification and a tinier
> fraction if the binary is smaller.

Yes, that was actually my point, that with larger file sizes and in real
mesh network scenario (multi-hop, not saturating the links) and more
realistic file sizes (say 200kB), the transfer will be taking significantly
more than the other operations.

Thanks!

--Simon

On Sat, Dec 7, 2019 at 5:20 AM Koen <koen@bergzand.net> wrote:

> Hi all,
> >     > [SS] Does that scale linearly with image size? Do memory
> requirements (RAM
> >     > usage) increase with the image size?
> >
> > I guess what we want to know is how much of that 7s time was the ED25519
> operation, and
> > how much time was the hash calculation.  Only the hash calculation would
> > scale with the image size.
>
> The time we measured spent on hashing (SHA256)is comparatively small for
> the typical firmware sizes.It takes ~4ms per kB to hashon the platform
> used in the paper. Even with a 200kB binary,it takes just a fraction
> (~10%) of the time needed for signature verification and a tinier
> fraction if the binary is smaller.
>
> Best regards,
> Koen Zandberg
>
> _______________________________________________
> Suit mailing list
> Suit@ietf.org
> https://www.ietf.org/mailman/listinfo/suit
>