[Mesa-dev] [PATCH] nir: add option to lower slt/sge/seq/sne

[Rob Clark]

On Tue, Mar 31, 2015 at 2:33 PM, Matt Turner <mattst88 at gmail.com> wrote:
> On Tue, Mar 31, 2015 at 11:30 AM, Rob Clark <robdclark at gmail.com> wrote:
>> On Tue, Mar 31, 2015 at 2:20 PM, Matt Turner <mattst88 at gmail.com> wrote:
>>> Alternatively, Jason's implemented a pass to "resolve" booleans to
>>> these values, which is necessary for us on i965 Gen4 and Gen5. You
>>> could probably use that.
>>
>> ok, is that already merged or on list somewhere?
>
> Yep, merged:
>

hmm, ok.. this looks like it could be more generic, but would mean
introducing variants of instructions returning bool which instead
returned int..

I guess we could just keep the mark-instructions-with-flags thing and
handle the flags in the driver backend..  which would be less
intrusive but it does seem nicer to have those other instructions
inserted before running other opt passes.

BR,
-R

> commit 41d64fa184671d372f6630deaf2401b00d4e984a
> Author: Jason Ekstrand <jason.ekstrand at intel.com>
> Date:   Tue Mar 17 11:49:04 2015 -0700
>
>     i965/nir: Do boolean resolves on GEN <= 5
>
>     v2: A couple comment clean-ups from Matt
>
>     Reviewed-by: Matt Turner <mattst88 at gmail.com>
>
> commit a55af2699feb8f64d6f480b223204da071606721
> Author: Jason Ekstrand <jason.ekstrand at intel.com>
> Date:   Tue Mar 17 11:29:01 2015 -0700
>
>     i965: Add a NIR analysis pass for determining when a boolean
> resolve is needed
>
>     v2: Fix the spelling of analyze and re-arrange code for better readability
>         as per Connor's comments.
>     v3: Make the naming of things more consistent and add a pile of comments
>     v4: Stop trying to avoid vectors
>
>     Reviewed-by: Matt Turner <mattst88 at gmail.com>
>     Reviewed-by: Connor Abbott <cwabbott0 at gmail.com>