harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Rana Dasgupta" <rdasg...@gmail.com>
Subject Re: [DRLVM][GC] (HARMONY-2398) patch for GCv5 alloc helper inlining
Date Thu, 07 Dec 2006 15:11:26 GMT
Mikhail,
   Thanks. Please see below...


On 12/7/06, Mikhail Fursov <mike.fursov@gmail.com> wrote:
>
> >Rana,
> >Allocation helper inlining is turned on only in server mode (-Xem:server)
> >and only for gc_cc


Yes, I realized after running the micro that the fast helper is checked in,
but it is not inlined yet :-) Is the server mode not the default mode? In
general, most common use case options need to be default, isn't it?

>Even running in server mode there is a nuance: we do not have OSR so let
> the
> >method to be recompiled before testing.


I did not understand this. Does this mean that after the initial compiles by
jet, we can come back later and recompile  hot methods with OPT, but we
cannot make inlining decisions without OSR? Why? Why won't the allocation
fastpath get inlined for the hot methods on the recompile?

I can help you to prepare configuration for -Xem:opt mode and rerun the
> test.


Could you please publish the necessary Xem:opt configuration here? Thanks.

>I think we will be as fast as SUN on windows, where FS[14] is used but
> >not system call to access TLS. The problem with TLS system call is quite
> >simple: we do not move it out from loop today.


Yes, I remember that we don't hoist the TLS call and VTune shows this to be
the hot method. We know this. But it would be good to know that this is the
only difference and not eg., nursery size etc.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message