[vlc-devel] [PATCH] arm_neon: Add an optimized routine for NV12/21 to I420

[Martin Storsjö]

On Mon, 30 Sep 2013, Denis Charmet wrote:

> Hi,
>
> Le lundi 30 septembre 2013 à 11:50:16, Martin Storsjö a écrit :
>> +	vld1.u8		{q0},		[IY,:128]!
>> +	pld		[UV, #16]
>> +	vld2.u8		{d2, d3},	[UV,:128]!
>
> <Noob question> why using only q0 and q1? isn't it possible to load Y in
> {q0,q1} for example? </noob question>

I guess that could be done as well, although it would increase some of the 
requirements for the alignment of the width. But I'm removing the copying 
of the Y data here anyway and replacing it with a plain memcpy as 
suggested by Rémi.

// Martin