[Cryptography] Speculation considered harmful but soft instruction sets not

John Levine johnl at iecc.com
Wed Jan 10 21:35:49 EST 2018


In article <20180110160247.GC10954 at localhost> you write:
>On Tue, Jan 09, 2018 at 05:46:07PM -0500, Theodore Ts'o wrote:
>VLIW == no stable ABI.  We'd have to ship bitcode and re-optimize,
>assemble, and link for every VLIW CPU model.
>
>It wouldn't be the end of the world.  But first we'd need the compilers
>that could handle it.

IBM has been doing this for decades on S/38 them AS/400 now called
something like Series i.  The underlying CPUs have completely changed,
but the binaries still work.  It's also how every Android phone works,
with the binaries in a java-like bytecode, and translated to native
form either when installed or when run.

I'm not saying this makes VLIW problems go away, but translating at
install time isn't a problem.

R's,
John


More information about the cryptography mailing list