News for hackers. Stuff that matters.
Re:Destroying x86_64 instruction decoders with differential ... by jcranmer on Thursday October 31, 2019 @12:03PM (#21410210)
The decoder doesn't actually take all that much space in the hardware, though. It's going to be smaller than the normal OoO logic, which means it's a pretty minor tax at best for actual hardware.
Re:Destroying x86_64 instruction decoders with differential ... by adwn on Thursday October 31, 2019 @03:30PM (#21412686)
> It's going to be smaller than the normal OoO logic, which means it's a pretty minor tax at best for actual hardware.

But it's a major tax for designing that hardware, and a potential source of bugs (the more complex, the more difficult to debug and verify).

Re:Destroying x86_64 instruction decoders with differential ... by darkmighty on Thursday October 31, 2019 @01:16PM (#21411200)
Instruction complexities can be used to save bandwidth/delay (and related energy consumption) at the cost of decoder size. Communication limitations are increasingly dominant in processors afaik; so the analysis is not so simple as to decoder size either.
Re:Destroying x86_64 instruction decoders with differential ... by jabl on Thursday October 31, 2019 @03:57PM (#21413040)
Compressed RISC ISA's (e.g. RISC-V C extension) achieve density comparable to x86, with a much simpler decoder.
Re:Destroying x86_64 instruction decoders with differential ... by baybal2 on Thursday October 31, 2019 @01:07PM (#21411058)
Saw die shots of Atom? Decoder makes more than half of the core, and there is no instruction cache.

In a loop, you will be spending more joules on decoding than actual computations.

Re:Destroying x86_64 instruction decoders with differential ... by monocasa on Thursday October 31, 2019 @02:02PM (#21411752)
What die shots are you looking at? I'm looking at a Silverthorne die and most of the frontend is taken up by the L1I$.

https://en.wikichip.org/wiki/intel/microarchitectures/bonnel...

Re:Destroying x86_64 instruction decoders with differential ... by baybal2 on Thursday October 31, 2019 @05:13PM (#21413765)
My bad