By: none (none.delete@this.none.com), June 30, 2013 5:03 am
Room: Moderated Discussions
EduardoS (no.delete@this.spam.com) on June 29, 2013 6:50 pm wrote:
[...]
> > Games being notable FP users.
>
> Not as much as you think.
Has there been a study about that? Both for desktop/console games and mobile games?
Anyway even if 30% of the instructions run are FP (which would be extremely high) and an OoO engine could double your speed (unlikely) you'd still only gain 15%.
Regarding Silvermont performance, I'm much more concerned by the 128+16 entries in the TLB, this is way too low, but I already said so :-)
[...]
> > Games being notable FP users.
>
> Not as much as you think.
Has there been a study about that? Both for desktop/console games and mobile games?
Anyway even if 30% of the instructions run are FP (which would be extremely high) and an OoO engine could double your speed (unlikely) you'd still only gain 15%.
Regarding Silvermont performance, I'm much more concerned by the 128+16 entries in the TLB, this is way too low, but I already said so :-)