Re: changing a collision resolution strategy of the symbol table of identifiers

[OndÅej BÃlka <neleai at seznam dot cz>]

On Sun, Oct 20, 2013 at 06:55:40PM +0600, Roman Gareev wrote:
> Dear gcc contributors,
> 
> Recently I came across the list of "ideas for speeding up GCC"
> (http://gcc.gnu.org/wiki/Speedup_areas). Among others, there was
> suggested to replace identifier hash table with other data structure.
> 
> Please find attached a patch, that switches the resolution strategy of
> a hash table used in the symbol table of identifiers from open
> addressing to separate chaining with linked list. I would be very
> grateful for your comments, feedback and ideas about further
> enhancements to gcc, in which I could take a part. I am pursuing
> master of science and looking for thesis theme, that is compiler
> related.
> 
> Below are results of testing and the patch.
> 
> During testing of the linux kernel (3.8.8) compilation time, the
> acquired results were the following: increasing by 0.17% for the
> version 4.8.0, increasing by 1.12% for the version 4.8.1, decreasing
> by 0.598% for trunk (this are average values).
> 
> 
> Tested x86_64-unknown-linux-gnu, applying to 4.8.0, 4.8.1 and trunk.
> 
> 
> diff --git a/gcc/stringpool.c b/gcc/stringpool.c
> 
> index f4d0dae..cc696f7 100644
> 
> --- a/gcc/stringpool.c
> 
> +++ b/gcc/stringpool.c
> 
> @@ -241,12 +241,8 @@ gt_pch_nx (unsigned char *x, gt_pointer_operator
> op, void *cookie)
> 
> /* SPD is saved in the PCH file and holds the information needed
> 
> to restore the string pool. */
> 
> -struct GTY(()) string_pool_data {
> 
> - ht_identifier_ptr *
> 
> - GTY((length ("%h.nslots"),
> 
Mail client broke it?