[Mesa-dev] [PATCH] ac/nir: mark some texture intrinsics as convergent

[Marek Olšák]

On Thu, May 30, 2019, 7:08 PM Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
wrote:

>
>
> On Fri, May 31, 2019, 12:49 AM Marek Olšák <maraeo at gmail.com> wrote:
>
>>
>>
>> On Thu, May 30, 2019, 6:44 PM Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
>> wrote:
>>
>>>
>>>
>>> On Thu, May 30, 2019, 11:45 PM Marek Olšák <maraeo at gmail.com> wrote:
>>>
>>>>
>>>>
>>>> On Thu, May 30, 2019, 3:54 PM Rhys Perry <pendingchaos02 at gmail.com>
>>>> wrote:
>>>>
>>>>> Otherwise LLVM can sink them and their texture coordinate calculations
>>>>> into divergent branches.
>>>>>
>>>>> v2: simplify the conditions on which the intrinsic is marked as
>>>>> convergent
>>>>>
>>>>> Cc: <mesa-stable at lists.freedesktop.org>
>>>>> Signed-off-by: Rhys Perry <pendingchaos02 at gmail.com>
>>>>> Reviewed-By: Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
>>>>> ---
>>>>>  src/amd/common/ac_nir_to_llvm.c | 12 ++++++++++++
>>>>>  1 file changed, 12 insertions(+)
>>>>>
>>>>> diff --git a/src/amd/common/ac_nir_to_llvm.c
>>>>> b/src/amd/common/ac_nir_to_llvm.c
>>>>> index 265e3b636c4..b1a191ac24c 100644
>>>>> --- a/src/amd/common/ac_nir_to_llvm.c
>>>>> +++ b/src/amd/common/ac_nir_to_llvm.c
>>>>> @@ -1394,6 +1394,18 @@ static LLVMValueRef build_tex_intrinsic(struct
>>>>> ac_nir_context *ctx,
>>>>>         }
>>>>>
>>>>>         args->attributes = AC_FUNC_ATTR_READNONE;
>>>>> +       /* Prevent texture instructions with implicit derivatives from
>>>>> being
>>>>> +        * sinked into branches. */
>>>>> +       switch (instr->op) {
>>>>> +       case nir_texop_tex:
>>>>> +       case nir_texop_txb:
>>>>> +       case nir_texop_lod:
>>>>> +               args->attributes |= AC_FUNC_ATTR_CONVERGENT;
>>>>> +               break;
>>>>> +       default:
>>>>> +               break;
>>>>> +       }
>>>>>
>>>>
>>>> I think this should only apply to the fragment shader.
>>>>
>>>
>>> Well these opcodes will only be used in fragment shaders anyway, right?
>>>
>>
>> I think the normal tex opcode can be used in vertex shaders as well,
>> probably only in OpenGL.
>>
>
> What is the behavior there? Implicit lod of 0?
>

Yes.

Marek


>> Marek
>>
>>
>>>> Marek
>>>>
>>>> +
>>>>>         return ac_build_image_opcode(&ctx->ac, args);
>>>>>  }
>>>>>
>>>>> --
>>>>> 2.21.0
>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20190530/a930f869/attachment.html>