News for hackers. Stuff that matters.
Re:Intel delays Cannonlake 10nm processor for the third time by throwaway2048 on Wednesday September 20, 2017 @01:33PM (#15296263)
snythetic benchmarks are notorious for being unrelistic and unreliable.

just because there isnt an alternative dosent mean that geekbench is accurate, id be pretty skeptical of claims that arm chips with a fraction of the transistors and power budget are somehow magically just as fast as big full fat x86 chips.

Re:Intel delays Cannonlake 10nm processor for the third time by robin_reala on Wednesday September 20, 2017 @04:03PM (#15297682)
Power delivery, sure, but you’re off on transistor count. Wikipedia list a ‘Dual-core + GPU Iris Core i7 Broadwell-U’ as 1.9B transistors [1], and the A11 is 4.3B.

[1] https://en.wikipedia.org/wiki/Transistor_count

Re:Intel delays Cannonlake 10nm processor for the third time by throwaway2048 on Thursday September 21, 2017 @01:23PM (#15305616)
Should be noted the A11 is 6cores+gpu and the i7u is 2 cores + GPU.

And we are talking about per core preformance.

Re:Intel delays Cannonlake 10nm processor for the third time by Dylan16807 on Wednesday September 20, 2017 @05:21PM (#15298397)
Current x86 chips are spending huge amounts of their transistor and power budget on tiny increases in instructions per clock.

When you have twice as many cores to play with then you can have better total performance with a lot less resource cost.