[Cryptography] Did Spectre help torpedo Qualcomm?

[Phillip Hallam-Baker]

On Sat, Sep 29, 2018 at 8:41 PM Florian Weimer <fw at deneb.enyo.de> wrote:

> * Henry Baker:
>
> > I wonder if the server chips Qualcomm developed were hopelessly
> > haunted by Spectre?
>
> I find that extremely unlikely.  There is a point (regarding cost or
> power efficiency) at which most customers would stop caring.  The
> whole zoo of issues is now largely perceived as a software problem
> anyway.
>

+1

There are a few applications and a few deployments in which SPECTRE might
lead to a compromise. Most of them are SSL private keys leaking in
co-located Web server environments.

As far as cryptography goes, just do all your private key operations on
trustworthy hardware, problem solved.

I suspect that a much bigger issue for Qualcom resulting from SPECTRE would
be that Intel suddenly required a large number of additional engineers and
were willing to almost pay anything to get them.

Before doing crypto, I did massively parallel computing. People should
probably start revisiting the Transputer architecture. Take a processor
with 16 cores, 16GB Ram and a 512GB CF Express card and you have a server
in a module the size of a 2.5" disk drive. Add a couple of 10GB Ethernet
ports and your only real problem is how to dispose of waste heat.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20180930/5ad0192d/attachment.html>