[PATCH 5/5] ia64: Use generic exp10f

Wilco Dijkstra Wilco.Dijkstra@arm.com
Thu Jun 18 17:18:04 GMT 2020


Hi Adhemerval,

On 09/04/2020 16:59, Adhemerval Zanella wrote:
> The generic implementation is slight worse (Itanium(R) Processor 9020):
> 
> Before new code:
>   "exp10f": {
>    "": {
>     "duration": 3.52876e+08,
>     "iterations": 2.1456e+07,
>     "max": 232.907,
>     "min": 7.766,
>     "mean": 16.4465
>    }
>   }
> 
> With new code:
>   "exp10f": {
>    "": {
>     "duration": 3.67279e+08,
>     "iterations": 2.1456e+07,
>     "max": 220.718,
>     "min": 15.503,
>     "mean": 17.1178
>    }
>   }
> 
> However it fixes all the issues on both:
> 
>   math/test-float-exp10
>   math/test-float32-exp10
> 
> (all the issues wrong results for non default rounding modes).

That's only a 4% slowdown - given it removes almost 600 lines of target
specific code and fixes various bugs, it's well worth it. LGTM.

Wilco


More information about the Libc-alpha mailing list