ReRe: performance (shellsort experiments)

[Tom Tromey]

>>>>> "Kresten" == Kresten Krab Thorup <krab@daimi.au.dk> writes:

Kresten>   static _Jv_VTable* __tmp = 0;
Kresten>   if (y->vtable != __tmp)
Kresten>     {
Kresten>        _Jv_CheckArrayStore (y, Foo.class); // may throw exc
Kresten>        __tmp = y->vtable;
Kresten>     }
Kresten>   x = y;   

The compiler would have to generate a lock of some sort here.  A spin
lock would be ok, maybe -- though _Jv_CheckArrayStore is a bit slow.

Kresten> On top of this, we can improve the performace of
Kresten> isAssignambleFrom; but I thing the most significant single
Kresten> thing one could do right now would be to implement uniquing
Kresten> for Utf8Const's, both in the linker and in the runtime
Kresten> system.

I'm curious to know why you picked this.

Tom