Running mpn_add_n at less than 1 cycle/limb

[Albin Ahlbäck]

On 30/08/2026 13:10, Torbjörn Granlund wrote:
> But why are we seeing things like
> https://gmplib.org/devel/tm/gmp/tuneup/success/gran.gmplib.org-stat:64.txt
> with lots of number < 1, then?
> 
> Because of deeo out-of-order execution, of course!  Two consecutive
> mpn_add_n invocations execute with some overlap.

Is `< 1' only applicable for throughput, or does it apply for latency as 
well?  I'm guessing that latency with less than 1 cycle per limb would 
require value prediction.

Best,
Albin