[Rose-public] Loop-invariant analysis

Tan Nguyen nnguyenthanh at eng.ucsd.edu
Wed Mar 18 17:23:25 PDT 2015


I just used the side effect analysis provided by ROSE to analyze whether
the value of a pointer can be potentially changed during the execution of
the enclosing loop. If anyone has another solution, please don't hesitate
to share it.

Thanks,
Tan.

On Wed, Mar 18, 2015 at 11:57 AM, Tan Nguyen <nnguyenthanh at eng.ucsd.edu>
wrote:

> Greg, I would like to replace pointer dereference expressions within a
> loop with array reference expressions. A simple case is when the value of a
> pointer does not changes throughout the execution of the enclosing loop, as
> shown in the following code snippet. Although I can write an analyzer for
> this case, it's worth learning how ROSE implemented a similar one (if any).
>
> double *p = &U[0];
> for(i=0; i<N; i++){
>   *(p+i) = 0.0;  => U[i] = 0.0;
> }
>
> Thanks,
> Tan.
>
> On Wed, Mar 18, 2015 at 8:30 AM, Bronevetsky, Greg <bronevetsky1 at llnl.gov>
> wrote:
>
>>  What kinds of invariants would you like to compute about loops?
>>
>>
>>
>> Greg Bronevetsky
>>
>> Lawrence Livermore National Lab
>>
>> Cell: (607) 279-3485
>>
>> bronevetsky at llnl.gov
>>
>> http://greg.bronevetsky.com
>>
>>
>>
>> *From:* rose-public-bounces at nersc.gov [mailto:
>> rose-public-bounces at nersc.gov] *On Behalf Of *Tan Nguyen
>> *Sent:* Tuesday, March 17, 2015 6:25 PM
>> *To:* rose-public at nersc.gov
>> *Subject:* [Rose-public] Loop-invariant analysis
>>
>>
>>
>> Hi, could someone tell me whether ROSE (or any project based on ROSE)
>> implements loop-invariant analysis? Thanks.
>>
>>
>>
>> --
>>
>> Thanks,
>>
>> Tan.
>>
>
>
>
> --
> Thanks,
> Tan.
>



-- 
Thanks,
Tan.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nersc.gov/pipermail/rose-public/attachments/20150318/9131322b/attachment.html>


More information about the rose-public mailing list