commit 47f01e538ad575d3f2b02e747905e7ab9dd3a9bb
author Jason Ekstrand <jason.ekstrand@intel.com> Thu May 05 17:14:39 2016 -0700
committer Jason Ekstrand <jason.ekstrand@intel.com> Wed May 11 11:44:35 2016 -0700
tree f5015584655d42317bcca2fce6b7c6078f5ccec1
parent 1b72c31e1f1947123d8c236b56e230f030f60cf9

ptn: Emit mul+add for MAD

Unlike fma() in GLSL, MAD in ARB programs is 100% splittable.  Just emit
the split version and let the optimizer fuse them later.

Shader-db results on Haswell:

   total instructions in shared programs: 7560379 -> 7560300 (-0.00%)
   instructions in affected programs: 143928 -> 143849 (-0.05%)
   helped: 443
   HURT: 250

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

src/mesa/program/prog_to_nir.c

diff --git a/src/mesa/program/prog_to_nir.c b/src/mesa/program/prog_to_nir.c
index b933553..9e01151 100644
--- a/src/mesa/program/prog_to_nir.c
+++ b/src/mesa/program/prog_to_nir.c
@@ -666,7 +666,7 @@
    [OPCODE_LIT] = 0,
    [OPCODE_LOG] = 0,
    [OPCODE_LRP] = 0,
-   [OPCODE_MAD] = nir_op_ffma,
+   [OPCODE_MAD] = 0,
    [OPCODE_MAX] = nir_op_fmax,
    [OPCODE_MIN] = nir_op_fmin,
    [OPCODE_MOV] = nir_op_fmov,
@@ -757,6 +757,10 @@
       ptn_lrp(b, dest, src);
       break;
 
+   case OPCODE_MAD:
+      ptn_move_dest(b, dest, nir_fadd(b, nir_fmul(b, src[0], src[1]), src[2]));
+      break;
+
    case OPCODE_DST:
       ptn_dst(b, dest, src);
       break;