This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][ipa-branch]MatrixTransposing Optimization


Hi,

this looks OK (at least from high level point of view).  Please guard
the verify_* calls with ENABLE_CHECKING...

Thanks,
Honza
> Hi,
> 
> 
> The performance measured on Linux PowerPC for matrix flattening on art 
> benchmark is 35% improvement, and 9% for equake benchmark. 
> If profiling is enabled, matrix transposing optimization can be 
> additionally performed. In this case, we get 2x improvement on art. 
> 
> O.K. for ipa branch?
> 
> Thanks,
> Razya
> 
> 2006-10-04  Razya Ladelsky  <razya@il.ibm.com>
> 
>         * matrix-reorg.c: (struct index_info) : Remove.
>         (struct access_site_info):  Add field 
> iterated_by_inner_most_loop_p
>         (struct matrix_info): Add dimension_size_orig.
>         Remove hottest_dim, reorg_failed_reason, flatten_p,
>         flatten_transpose_matrix_p,  flatten_decl.
>         (analyze_transpose, mat_free,  sort_dim_hot_level ): New function.
>         (check_transpose_p, analyze_transpose_p) : New static variables.
>         (record_access_alloc_site_info, check_allocation_function,
>         analyze_matrix_accesses, find_escaping_matrices_and_flatten,
>         transform_access_sites, matrix_reorg):  Add support for 
> transposing.
> 
> 
> 
> 









Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]