>
> Hi Dave,
>
> Am 07.11.2010 um 02:47 schrieb [hidden email]:
>> Back when I did Berkeley Smalltalk (early 1980's), I fretted a lot about incremental performance improvements to the interpreter.
>> Every few days I was able to get another few percent, as I recall. Then Peter and Alan did the first OO JIT ([Deutsch-Schiffman '84])
>> and it was so much faster than BS, it became clear to me that I didn't want to spend any more of my time optimizing an interpreter
>> when I could be adding a simple compiler instead. But that's just own experience.
>
> sure! Those interpreter considerations are just one part of the slides ... JIT compilers are there as well.
>
> One interesting observation that students made with their threading implementations is that it did not significantly improve performance: today's branch predictors seem to be too advanced to be impressed by threading. (They measured this using HPM.) Of course, the VM used in that course is really simple, and also has a very simple and small bytecode set.
>
> Best,
>
> Michael
>
>>