[Mesa-dev] [PATCH 03/20] nir: Support lowering vote intrinsics

[Matt Turner]

On Thu, Jul 6, 2017 at 8:04 PM, Connor Abbott <cwabbott0 at gmail.com> wrote:
> On Thu, Jul 6, 2017 at 4:48 PM, Matt Turner <mattst88 at gmail.com> wrote:
>> ... trivially (as allowed by the spec!) by reusing the existing
>> nir_opt_intrinsics code.
>> ---
>>  src/compiler/nir/nir.h                | 4 ++++
>>  src/compiler/nir/nir_opt_intrinsics.c | 6 +++---
>>  2 files changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
>> index 44a1d0887e..401c41f155 100644
>> --- a/src/compiler/nir/nir.h
>> +++ b/src/compiler/nir/nir.h
>> @@ -1821,6 +1821,10 @@ typedef struct nir_shader_compiler_options {
>>     bool lower_extract_byte;
>>     bool lower_extract_word;
>>
>> +   bool lower_vote_any;
>> +   bool lower_vote_all;
>> +   bool lower_vote_eq;
>
> Since there are potentially multiple ways to lower these (voteAny(x)
> -> !voteAll(!x), using ballotARB(), etc.), and the way they're lowered
> is a little... unexpected (although admittedly legal!), why don't we
> use a more descriptive name, like lower_vote_*_trivial? While we're at
> it, I highly doubt that an implementation would want this kind of
> lowering for just one of the intrinsics, so we can merge this into a
> single flag, say lower_vote_trivial.

Thanks, both good ideas. I've replaced all three fields with a
lower_vote_trivial field.