| setsunatenshi said: if this rumour turns out to be true then brace for disappointment in many fronts, processing power, energy efficiency and even cost. older chips in bigger sizes don't go down in price as time goes by so much as using current tech. bigger sized chip means higher TDP which is killer for a handheld, so expect the battery not lasting as long as it would under Pascal. |
torok said:
It's not, they are wrong. The Pascal variant wouldn't reach X1, the Maxwell one will be closer to half X1's power. I don't like to use flops as argument, but Pascal Tegra is 750 Gflops and Maxwell Tegra is 512 Gflops. When translating this to X1 equivalent performance, it's probably even weaker. I don't think this will have a lot of influence in how it succeds. The 3rd party support would be limited anyway, better to go with the cheaper SoC than try to make publishers happy. Just go for an aggressive 175 bucks price point. It's feasible. |
It's like you didnt' read the post right above yours posted an hour before...
Using Pascal and newer fab architecture is not about beating X1 or pulling off miracles of performance.
But not using it needlessly gimps performance and will not achieve long term cost benefit.
The primary historic purpose of fab node progression is cost reduction after all.
Launching with an obsolete fab node means they will have to pay for serious redesign at later point,
and they lock themselves out of benefits of designing around benefits of modern fab node in the first place.
The redesign costs are serious because it would not just be a simple shrink but move from 2d to 3d design.
If they go with 14nm or 10nm now, then moving to 7m 3d does not require redesign like 2d to 3d.
