[Mesa-dev] [PATCH v2 39/42] i965/nir: Implement shared variable atomic operations

[Iago Toral]

On Tue, 2015-11-17 at 21:55 -0800, Jordan Justen wrote:
> Signed-off-by: Jordan Justen <jordan.l.justen at intel.com>
> ---
>  src/mesa/drivers/dri/i965/brw_fs.h       |  2 ++
>  src/mesa/drivers/dri/i965/brw_fs_nir.cpp | 60 ++++++++++++++++++++++++++++++++
>  2 files changed, 62 insertions(+)
> 
> diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h
> index cbfc07f..1c9b4c3 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs.h
> +++ b/src/mesa/drivers/dri/i965/brw_fs.h
> @@ -276,6 +276,8 @@ public:
>                             nir_intrinsic_instr *instr);
>     void nir_emit_ssbo_atomic(const brw::fs_builder &bld,
>                               int op, nir_intrinsic_instr *instr);
> +   void nir_emit_shared_atomic(const brw::fs_builder &bld,
> +                               int op, nir_intrinsic_instr *instr);
>     void nir_emit_texture(const brw::fs_builder &bld,
>                           nir_tex_instr *instr);
>     void nir_emit_jump(const brw::fs_builder &bld,
> diff --git a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> index c8c6370..792fda7 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> @@ -1967,6 +1967,37 @@ fs_visitor::nir_emit_cs_intrinsic(const fs_builder &bld,
>        break;
>     }
>  
> +   case nir_intrinsic_shared_atomic_add:
> +      nir_emit_shared_atomic(bld, BRW_AOP_ADD, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_min:
> +      if (dest.type == BRW_REGISTER_TYPE_D)
> +         nir_emit_shared_atomic(bld, BRW_AOP_IMIN, instr);
> +      else
> +         nir_emit_shared_atomic(bld, BRW_AOP_UMIN, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_max:
> +      if (dest.type == BRW_REGISTER_TYPE_D)
> +         nir_emit_shared_atomic(bld, BRW_AOP_IMAX, instr);
> +      else
> +         nir_emit_shared_atomic(bld, BRW_AOP_UMAX, instr);
> +      break;

As explained in the previous patch, this implementation of min/max won't
work. Check the ssbo implementation in master.

> +   case nir_intrinsic_shared_atomic_and:
> +      nir_emit_shared_atomic(bld, BRW_AOP_AND, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_or:
> +      nir_emit_shared_atomic(bld, BRW_AOP_OR, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_xor:
> +      nir_emit_shared_atomic(bld, BRW_AOP_XOR, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_exchange:
> +      nir_emit_shared_atomic(bld, BRW_AOP_MOV, instr);
> +      break;
> +   case nir_intrinsic_shared_atomic_comp_swap:
> +      nir_emit_shared_atomic(bld, BRW_AOP_CMPWR, instr);
> +      break;
> +
>     default:
>        nir_emit_intrinsic(bld, instr);
>        break;
> @@ -2607,6 +2638,35 @@ fs_visitor::nir_emit_ssbo_atomic(const fs_builder &bld,
>  }
>  
>  void
> +fs_visitor::nir_emit_shared_atomic(const fs_builder &bld,
> +                                   int op, nir_intrinsic_instr *instr)
> +{
> +   fs_reg dest;
> +   if (nir_intrinsic_infos[instr->intrinsic].has_dest)
> +      dest = get_nir_dest(instr->dest);
> +
> +   unsigned index = BRW_SLM_SURFACE_INDEX;
> +   fs_reg surface = fs_reg(index);

fs_reg surface = brw_imm_ud(BRW_SLM_SURFACE_INDEX) and remove the index
variable.

Iago

> +   fs_reg offset = get_nir_src(instr->src[0]);
> +   fs_reg data1 = get_nir_src(instr->src[1]);
> +   fs_reg data2;
> +   if (op == BRW_AOP_CMPWR)
> +      data2 = get_nir_src(instr->src[2]);
> +
> +   /* Emit the actual atomic operation operation */
> +
> +   fs_reg atomic_result =
> +      surface_access::emit_untyped_atomic(bld, surface, offset,
> +                                          data1, data2,
> +                                          1 /* dims */, 1 /* rsize */,
> +                                          op,
> +                                          BRW_PREDICATE_NONE);
> +   dest.type = atomic_result.type;
> +   bld.MOV(dest, atomic_result);
> +}
> +
> +void
>  fs_visitor::nir_emit_texture(const fs_builder &bld, nir_tex_instr *instr)
>  {
>     unsigned sampler = instr->sampler_index;