[PING][PATCHv3 1/2] aarch64: Hoist ZVA check out of the memset function

[Siddhesh Poyarekar]

On Monday 16 October 2017 06:37 PM, Szabolcs Nagy wrote:
> i think the patch can be reduced to have
> 3 variants instead of 4 (the nozva case does
> not seem to buy much over the generic code)

How about this then for least change; I'll just add an alternate routine
for zva == 64 and keep everything else the same.

> and the alignments have to be checked for the
> new fixed zva size cases because they seem to
> be wrong.

I'll keep the routine separate so that it's easier for you to see the
alignments of various bits.

> it's not entirely clear what should happen
> for libc internal memset calls, that needs
> more thought.

Default to the current memset?

> quick test shows that the current benchmark was
> not running enough iterations and not testing
> enough cases, i'd expect the actual speed up
> to be lower than previously reported.

It's running enough iterations, it's just not running across the array
like I had planned, so the iterations are overlapping and hence
ineffective.  I'll post a fix and also the new numbers.

Siddhesh