Wilco (Wilco.Dijkstra.delete@this.ntlworld.com) on November 21, 2012 6:41 am wrote:
> David Kanter (dkanter.delete@this.realworldtech.com) on November 20, 2012 11:07 pm wrote:
> > Rohit (a.delete@this.b.c) on November 20, 2012 4:48 pm wrote:
> > > David Kanter (dkanter.delete@this.realworldtech.com) on November 19, 2012 3:40 pm wrote:
> > > > That's probably a common view. However, there are a number of optimizations that can net
> > > > you huge performance gains and are only viable for JIT systems. I agree you want to do
> > > > as much as possible in your compiler, but somethings require dynamic information.
> > > >
> > > > DK
> > >
> > > Which optimizations are these?
> >
> > Speculative multi-threading using a JIT.
>
> That's not at all JIT related. Has anyone ever shown a speedup using software-only speculation?

Yes, albeit only small ones.

> Automatic multihreading ARMs have been studied for many years. However even with special hardware
> support to detect conflicts, fast rollbacks and spawning threads to different cores it never
> seemed worth the relatively small speedups, especially not when you consider the extra power
> for speculating. The big speedups were all in loops which can be dealt with normally.

Apparently they were studied poorly. I can tell you from my own experience that ARM really has minimal interest in this area and has explored it far less than other CPU companies.

You don't really pay extra power for speculating. If you compare two well-designed systems; one can run the same software on 2 threads and one on 4 threads...there's going to be quite a performance gap. So as long as your speculative overhead isn't 50%, you're probably OK.

Mis-speculation can consume quite a bit of power, but that just means you have to be careful.

> To repeat, I don't believe there are any optimizations which are much better done in a JIT than in
> a static compiler. Even dynamic info like profiling works better when done the >traditional way.

You're welcome to believe that, but you'd be wrong. I've seen plenty of data to the contrary.

Moreover, there's plenty of public data from TMTA about optimizations which are only possible through JITing.

DK