[Mesa-dev] [PATCH v2 11/30] i965/fs: add shuffle_32bit_load_result_to_64bit_data helper

[Iago Toral]

On Thu, 2016-05-12 at 20:01 -0700, Francisco Jerez wrote:
> Samuel Iglesias Gonsálvez <siglesias at igalia.com> writes:
> 
> > From: Iago Toral Quiroga <itoral at igalia.com>
> >
> > There will be a few places where we need to shuffle the result of a 32-bit
> > load into valid 64-bit data, so extract this logic into a separate helper
> > that we can reuse.
> >
> > The shuffling needs to operate with WE_all set because we are changing the
> > layout of the data across the various channels. Otherwise we will run into
> > problems in non-uniform control-flow scenarios.
> >
> I guess you could remove this paragraph because it no longer applies to
> the last version of the patch.

Ooops, yes!

> > v2 (Curro):
> > - Use subscript() instead of stride()
> > - Assert on the input types rather than retyping.
> > - Use offset() instead of horiz_offset(), drop the multiplier definition.
> > - Do not use a temporary for the writes and drop force_writemask_all.
> 
> Don't pretend you took my "don't use a temporary" suggestion into
> account. :P

Oh right, I did write the patch with that change included but then run
into the issue we discussed when src == dst and forgot to remove this. 

> > - Mark component_i as const.
> 
> Did you forget to git-add something?

No, somehow I lost the change in the process... I'll put it back.

> > - Make the function name lower case.
> > ---
> >  src/mesa/drivers/dri/i965/brw_fs.cpp | 53 ++++++++++++++++++++++++++++++++++++
> >  src/mesa/drivers/dri/i965/brw_fs.h   |  5 ++++
> >  2 files changed, 58 insertions(+)
> >
> > diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
> > index 15d5759..4827dea 100644
> > --- a/src/mesa/drivers/dri/i965/brw_fs.cpp
> > +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
> > @@ -212,6 +212,59 @@ fs_visitor::VARYING_PULL_CONSTANT_LOAD(const fs_builder &bld,
> >  }
> >  
> >  /**
> > + * This helper takes the result of a load operation that reads 32-bit elements
> > + * in this format:
> > + *
> > + * x x x x x x x x
> > + * y y y y y y y y
> > + * z z z z z z z z
> > + * w w w w w w w w
> > + *
> > + * and shuffles the data to get this:
> > + *
> > + * x y x y x y x y
> > + * x y x y x y x y
> > + * z w z w z w z w
> > + * z w z w z w z w
> > + *
> > + * Which is exactly what we want if the load is reading 64-bit components
> > + * like doubles, where x represents the low 32-bit of the x double component
> > + * and y represents the high 32-bit of the x double component (likewise with
> > + * z and w for double component y). The parameter @components represents
> > + * the number of 64-bit components present in @src. This would typically be
> > + * 2 at most, since we can only fit 2 double elements in the result of a
> > + * vec4 load.
> > + *
> > + * Notice that @dst and @src can be the same register.
> > + */
> > +void
> > +fs_visitor::shuffle_32bit_load_result_to_64bit_data(const fs_builder &bld,
> 
> There was a second reason I had in mind when I suggested it would
> improve encapsulation to take this out of fs_visitor: The function has
> absolutely nothing to do with visiting, it uses no internal or external
> fs_visitor data structures or interfaces, it doesn't even use the "this"
> pointer.  Defining a function that could perfectly be stand-alone inside
> an object it has no need to be in (in this case it doesn't even have any
> logical relation with) actually *decreases* encapsulation because it
> exposes the object's internals to the function unnecessarily.
> 
> Either way the back-end code is already plagued by this anti-pattern so
> I'm not going to complain if you keep the code as-is -- You could argue
> you're just being consistent with the existing practice. ;)

I agree with all your reasoning, my only disagreement is with the fact
that brw_fs_nir.cpp is a better place for it considering that this has
nothing to do with NIR, but since we have to put this somewhere let's
put it there for now.

> > +                                                    const fs_reg dst,
> > +                                                    const fs_reg src,
> 
> Pass by reference.

Ok.

> > +                                                    uint32_t components)
> > +{
> > +   assert(type_sz(src.type) == 4);
> > +   assert(type_sz(dst.type) == 8);
> > +
> > +   /* A temporary that we will use to shuffle the 32-bit data of each
> > +    * component in the vector into valid 64-bit data. We can't write directly
> > +    * to dst because dst can be (and would usually be) the same as src
> > +    * and in that case the first MOV in the loop below would overwrite the
> > +    * data read in the second MOV.
> > +    */
> > +   fs_reg tmp = bld.vgrf(dst.type);
> > +
> > +   for (unsigned i = 0; i < components; i++) {
> > +      fs_reg component_i = offset(src, bld, 2 * i);
> > +
> > +      bld.MOV(subscript(tmp, src.type, 0), component_i);
> > +      bld.MOV(subscript(tmp, src.type, 1), offset(component_i, bld, 1));
> > +
> > +      bld.MOV(offset(dst, bld, i), tmp);
> 
> Ah, yes, this looks much better,

Yes, I agree :)

> Reviewed-by: Francisco Jerez <currojerez at riseup.net>
> 
> > +   }
> > +}
> > +
> > +/**
> >   * A helper for MOV generation for fixing up broken hardware SEND dependency
> >   * handling.
> >   */
> > diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h
> > index b906f3d..1970ad0 100644
> > --- a/src/mesa/drivers/dri/i965/brw_fs.h
> > +++ b/src/mesa/drivers/dri/i965/brw_fs.h
> > @@ -105,6 +105,11 @@ public:
> >                                     uint32_t const_offset);
> >     void DEP_RESOLVE_MOV(const brw::fs_builder &bld, int grf);
> >  
> > +   void shuffle_32bit_load_result_to_64bit_data(const brw::fs_builder &bld,
> > +                                                const fs_reg dst,
> > +                                                const fs_reg src,
> > +                                                uint32_t components);
> > +
> >     bool run_fs(bool do_rep_send);
> >     bool run_vs(gl_clip_plane *clip_planes);
> >     bool run_tcs_single_patch();
> > -- 
> > 2.5.0
> >
> > _______________________________________________
> > mesa-dev mailing list
> > mesa-dev at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev