[Cryptography] RISC-V isn't the answer

[Dan McDonald]

On Mon, Jan 22, 2018 at 04:32:29PM -0800, Tony Arcieri wrote:
> On Mon, Jan 22, 2018 at 2:34 PM, Bill Frantz <frantz at pwpconsult.com> wrote:
> > One approach I like is a massive number of simple cores that don't
> > speculate.
> 
> 
>  This approach is where RISC-V soars today:
> https://fuse.wikichip.org/news/686/esperanto-exits-stealth-mode-aims-at-ai-with-a-4096-core-7nm-risc-v-monster/
> 

I hope appropriate workloads find this beast helpful.  I remember the Afara
Websystems SPARC design, which was bought by Sun and became the Niagara
family of processors (T1000, T2000, etc. etc.).  Niagara did poorly on more
traditional workloads, and won market share more through energy efficiency
than raw performance.

Maybe with side-channels becoming more of a not-just-us-security-types
problem, such massively-cored architectures can have another go.

Dan