--- Comment #49 from 2010-07-25 17:21:10 PDT ---

Thanks for doing these tests!

In D1, the runtime calls gc_query on _every_ array append operation. Though the
cache in Gcx.getInfo() speeds it up a bit. But it doesn't look like
gc_query/getInfo() needs to do more work because of this patch?

You also could try to replace the code in the mark function by the old,
non-precise one. This way you can see how much overhead the bitmask
manipulation code causes.

Configure issuemail:
------- You are receiving this mail because: -------

Reply via email to