Something made the Sunspider time for JM+TI on awfy go from ~300ms to ~350ms. Expected?
It might be related to the diagnostic patch from Bill that adds a memset in the object-finalization path.
memset(t, JS_FREE_PATTERN, sizeof(T));
This regression came in on a merge from m-c (it would be on TM too, but TM hasn't pulled from m-c for a couple weeks). It is fixed by bug 672283, I just haven't pulled that fix to JM yet (should be able to do this later today).