Re: [Jikesrvm-researchers] Rapid Type Analysis in Jikes RVM

[David W. <cw...@ui...>]

Hi Feng,

Since I am using RTA to simplify my call graph in my interprocedural 
analysis, I am interested to know whether it can reduce the number of 
candidate methods given a virtual/interface method call.  It does not 
have to be a singleton set (as used by static inlining).  Also, does RTA 
increase the number of method invalidations when compared with CHA?

Thanks for the information,

david

Feng Qian wrote:

> Hi,
>
> I had an implementation of dynamic RTA in my private working 
> directory. So far, I didnot see improvement over CHA (for static 
> inlining) although I am looking at it more carefully. I am willing to 
> share information with you if you are interested in.
>
> Cheers,
> Feng
>
> wong chi-leung wrote:
>
>> Hello,
>>
>> It seems that Jikes RVM performs static inlining based on class
>> hierarchy analysis.  Is there any intrinsic reason why rapid type
>> analysis is not used (e.g. by looking at all bytecode to find out the
>> set of classes participating in "new" bytecode)?
>>
>> Thanks for the information,
>>
>> david
>> _______________________________________________
>> Jikesrvm-researchers mailing list
>> Jik...@os...
>> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jikesrvm-researchers 
>>
>>  
>>
>
> _______________________________________________
> Jikesrvm-researchers mailing list
> Jik...@os...
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jikesrvm-researchers 
>