[PATCH v2 2/5] benchtests: Add memset zero fill benchtest

[naohirot@fujitsu.com]

Hi Wilco,

> 
>       TIMING_NOW (start);
>       for (j = 0; j < 16; j++)
>         CALL (impl, s + n * j, c2, n);
>       TIMING_NOW (stop);
>  
> This loop is basically equivalent to CALL (impl, s, c2, n * 16), so you

Yes, but the number of function call is different between 1 time and
16 times.

> might as well change the outer loop to use a larger 'n'. The accuracy
> will be bad unless 'n' is really large, and there is no way to improve it.

Umm I couldn't understand the logic of this part.
How do we change the the outer loop to use a larger 'n'?

> If you want to test zero/non-zero combinations accurately, store the
> memset values in an array rather than using a single constant.

I changed one line as shown in the mail [1]
from
  CALL (impl, s, c1, n * 16);
to
  __builtin_memset (s, c1, n * 16);

Is this the array you mentioned?

[1] https://sourceware.org/pipermail/libc-alpha/2021-July/129461.html

Thanks.
Naohiro