Hi,

Sorry, please ignore the previously attached file, which isn't the latest one
although almost are the same.   The latest tested is attached here.  

Sorry for the inconvenience.

BR,
Kewen

on 2020/7/22 下午11:48, Kewen.Lin via Gcc-patches wrote:
> 
> It's a great idea, by following your subsequent suggestion to make the structure
> like:
> 
>   - calculate peel_iters_prologue
>   - calculate peel_iters_epilogue
>   - add costs associated with peel_iters_prologue
>   - add costs associated with peel_iters_epilogue
>   - add costs related to branch taken/not_taken.
> 
> the updated v3 is attached.
> 
> Just bootstrapped/regtested on powerpc64le-linux-gnu (P9) with explicit
> param vect-partial-vector-usage=1, I'll test it without partial vectors
> setting, also on aarch64 later.
> 
> BR,
> Kewen
> -----
> gcc/ChangeLog:
> 
> 	* config/rs6000/rs6000.c (adjust_vect_cost_per_loop): New function.
> 	(rs6000_finish_cost): Call adjust_vect_cost_per_loop.
> 	* tree-vect-loop.c (vect_get_known_peeling_cost): Factor out some code
> 	to determine peel_iters_epilogue to function ...
> 	(vect_get_peel_iters_epilogue): ... this.  New function.
> 	(vect_estimate_min_profitable_iters):  Add cost modeling for vector
> 	with length, refactor cost calculation on peel_iters_prologue and
> 	peel_iters_epilogue.
>