[Qemu-devel] Why some ARM NEON helper functions need mask?

[Qemu-devel] Why some ARM NEON helper functions need mask?

[陳韋任]

Hi, all

  I am looking into QEMU's implementation for ARM NEON instructions
(target-arm/neon_helper.c). Some helper functions will do mask
operation, neon_add_u8, for example. I thought simply adding a and b
is enough and can't figure out why the mask operation is needed.

---
uint32_t HELPER(neon_add_u8)(uint32_t a, uint32_t b)
{
    uint32_t mask;
    mask = (a ^ b) & 0x80808080u;
    a &= ~0x80808080u;
    b &= ~0x80808080u;
    return (a + b) ^ mask;
}
---

  Any help is appreciated.

Regards,
chenwj

-- 
Wei-Ren Chen (陳韋任)
Computer Systems Lab, Institute of Information Science,
Academia Sinica, Taiwan (R.O.C.)
Tel:886-2-2788-3799 #1667

Re: [Qemu-devel] Why some ARM NEON helper functions need mask?

[Max Filippov]

>   I am looking into QEMU's implementation for ARM NEON instructions
> (target-arm/neon_helper.c). Some helper functions will do mask
> operation, neon_add_u8, for example. I thought simply adding a and b
> is enough and can't figure out why the mask operation is needed.

These are SIMD instructions acting upon independent data 'lanes' packed into bigger data item.
Lane operations must not interfere with each other.
 
> ---
> uint32_t HELPER(neon_add_u8)(uint32_t a, uint32_t b)
> {
>     uint32_t mask;
>1:     mask = (a ^ b) & 0x80808080u;
>2:     a &= ~0x80808080u;
>3:     b &= ~0x80808080u;
>4:     return (a + b) ^ mask;
> }
> ---

In your example there are four 8-bit lanes packed into 32-bit word.
If we add whole 32-bit words then care must be taken to prevent overflow propagation between the lanes.
This is done by putting zero at the top bit of each 8-bit operand (steps 2 and 3).
These top bits are summed modulo 2 separately (step 1) and then added back (step4).

Thanks.
-- Max

Re: [Qemu-devel] Why some ARM NEON helper functions need mask?

[Chih-Min Chao]

On Sun, Oct 30, 2011 at 7:39 PM, 陳韋任 <chenwj@iis.sinica.edu.tw> wrote:
> Hi, all
>
>  I am looking into QEMU's implementation for ARM NEON instructions
> (target-arm/neon_helper.c). Some helper functions will do mask
> operation, neon_add_u8, for example. I thought simply adding a and b
> is enough and can't figure out why the mask operation is needed.
>
> ---
> uint32_t HELPER(neon_add_u8)(uint32_t a, uint32_t b)
> {
>    uint32_t mask;
>    mask = (a ^ b) & 0x80808080u;
>    a &= ~0x80808080u;
>    b &= ~0x80808080u;
>    return (a + b) ^ mask;
> }
> ---
>
ex :

a =  0x01  01 01 01
b = 0xFF  FF FF FF

the expected result of a+ b is
0x0   0   0   0

simply add a to b is
0x1   1   1   0

>  Any help is appreciated.
>
> Regards,
> chenwj
>
> --
> Wei-Ren Chen (陳韋任)
> Computer Systems Lab, Institute of Information Science,
> Academia Sinica, Taiwan (R.O.C.)
> Tel:886-2-2788-3799 #1667
>
>

Re: [Qemu-devel] Why some ARM NEON helper functions need mask?

[陳韋任]

  Thanks, Max and Chih-Min. It's much clear to me now.

Regards,
chenwj

-- 
Wei-Ren Chen (陳韋任)
Computer Systems Lab, Institute of Information Science,
Academia Sinica, Taiwan (R.O.C.)
Tel:886-2-2788-3799 #1667