[PATCH v2 01/25] benchtests: Add acosf benchmark
Adhemerval Zanella Netto
adhemerval.zanella@linaro.org
Wed Dec 18 13:05:47 GMT 2024
On 06/12/24 20:28, DJ Delorie wrote:
> Adhemerval Zanella <adhemerval.zanella@linaro.org> writes:
>> The input is based on acosf one (random inputs in [-1,1]).
>
> What? This *is* the acosf one. And we already have acosf-inputs in
> benchtests/libmvec/ but this doesn't match that. Can we merge them
> somehow?
Oops, it should be 'acos' and it is based in the sense it is random
values in the range og [-1,-1].
>
> LGTM otherwise
> Reviewed-by: DJ Delorie <dj@redhat.com>
Thanks.
>
>> diff --git a/benchtests/Makefile b/benchtests/Makefile
>> index 2ee5516d55..924434903a 100644
>> --- a/benchtests/Makefile
>> +++ b/benchtests/Makefile
>> @@ -25,6 +25,7 @@ include ../Makeconfig
>>
>> bench-math := \
>> acos \
>> + acosf \
>
> Ok.
>
>> diff --git a/benchtests/acosf-inputs b/benchtests/acosf-inputs
>> new file mode 100644
>> index 0000000000..bda4c44162
>> --- /dev/null
>> +++ b/benchtests/acosf-inputs
>> @@ -0,0 +1,2710 @@
>> +## args: float
>> +## ret: float
>> +## includes: math.h
>> +# Random inputs in [-1,1]
>> +## name: workload-core-math1
>
> Ok.
>
>> +0x1.8c9cfp-3
>> +0x1.c8d8bp-1
>> +-0x1.58bbe8p-2
>> . . .
>> +0x1.b0384cp-1
>> +0x1.3e9f5p-2
>> +-0x1.00d68p-2
>
> Ok.
>
More information about the Libc-alpha
mailing list