[Mesa-dev] [PATCH 5/5] nir: Recognize open-coded bitfield_reverse.

[Jason Ekstrand]

On Thu, Feb 4, 2016 at 5:48 PM, Matt Turner <mattst88 at gmail.com> wrote:

> Helps 11 shaders in UnrealEngine4 demos.
>
> I seriously hope they would have given us bitfieldReverse() if we
> exposed GL 4.0 (but we do expose ARB_gpu_shader5, so why not use that
> anyway?).
>
> instructions in affected programs: 4875 -> 4633 (-4.96%)
> cycles in affected programs: 270516 -> 244516 (-9.61%)
>
> I suspect there's a *lot* of room to improve nir_search/opt_algebraic's
> handling of this. We'd actually like to match, e.g., step2 by matching
> step1 once and then doing a pointer comparison for the second instance
> of step1, but unfortunately we generate an enormous tuple for instead.
>
> The .text size increases by 6.5% and the .data by 17.5%.
>
>    text     data  bss    dec    hex  filename
>   22957    45224    0  68181  10a55  nir_libnir_la-nir_opt_algebraic.o
>   24461    53160    0  77621  12f35  nir_libnir_la-nir_opt_algebraic.o
>
> I'd be happy to remove this if Unreal4 uses bitfieldReverse() if it is
> in a GL 4.0 context once we expose GL 4.0.
> ---
> Maybe it'd be better do make this a separate pass capable of recognizing
> this
> pattern without blowing up the compiled code size. Probably worth checking
> whether they use bitfieldReverse() under GL 4.0 first...
>
>  src/compiler/nir/nir_opt_algebraic.py | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
>
> diff --git a/src/compiler/nir/nir_opt_algebraic.py
> b/src/compiler/nir/nir_opt_algebraic.py
> index 0a248a2..f92c6b9 100644
> --- a/src/compiler/nir/nir_opt_algebraic.py
> +++ b/src/compiler/nir/nir_opt_algebraic.py
> @@ -311,6 +311,18 @@ optimizations = [
>       'options->lower_unpack_snorm_4x8'),
>  ]
>
> +def bitfield_reverse(u):
> +    step1 = ('ior', ('ishl', u, 16), ('ushr', u, 16))
> +    step2 = ('ior', ('ishl', ('iand', step1, 0x00ff00ff), 8), ('ushr',
> ('iand', step1, 0xff00ff00), 8))
> +    step3 = ('ior', ('ishl', ('iand', step2, 0x0f0f0f0f), 4), ('ushr',
> ('iand', step2, 0xf0f0f0f0), 4))
> +    step4 = ('ior', ('ishl', ('iand', step3, 0x33333333), 2), ('ushr',
> ('iand', step3, 0xcccccccc), 2))
> +    step5 = ('ior', ('ishl', ('iand', step4, 0x55555555), 1), ('ushr',
> ('iand', step4, 0xaaaaaaaa), 1))
> +
> +    return step5
>

Mind calling this "ue4_bitfield_reverse"?  You're not detecting a generic
bitfield reverse here.  With that, patches 1, 3, and 5 are

Reviewed-by: Jason Ekstrand <jason.ekstrand at intel.com>

I trust Dylan on patch 4.  I was just trying to ensure that we got/used a
32-bit value.
--Jason


> +
> +optimizations += [(bitfield_reverse('x'), ('bitfield_reverse', 'x'))]
> +
> +
>  # Add optimizations to handle the case where the result of a ternary is
>  # compared to a constant.  This way we can take things like
>  #
> --
> 2.4.10
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20160208/e9d1e454/attachment.html>