From 715f0d06d19e7c33d98f99c764c5c3249d13b1c0 Mon Sep 17 00:00:00 2001
From: Timothy Arceri <timothy.arceri@collabora.com>
Date: Tue, 13 Dec 2016 11:36:51 +1100
Subject: [PATCH] i965: use nir loop unrolling pass

shader-db results for BDW:

total instructions in shared programs: 12589614 -> 12590119 (0.00%)
instructions in affected programs: 50525 -> 51030 (1.00%)
helped: 7
HURT: 145

total cycles in shared programs: 241524604 -> 241490502 (-0.01%)
cycles in affected programs: 1941404 -> 1907302 (-1.76%)
helped: 302
HURT: 449

total loops in shared programs: 4245 -> 2947 (-30.58%)
loops in affected programs: 1535 -> 237 (-84.56%)
helped: 1142
HURT: 0

total spills in shared programs: 14453 -> 14453 (0.00%)
spills in affected programs: 0 -> 0
helped: 0
HURT: 0

total fills in shared programs: 18984 -> 18984 (0.00%)
fills in affected programs: 0 -> 0
helped: 0
HURT: 0

LOST:   26
GAINED: 15

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
---
 src/mesa/drivers/dri/i965/brw_compiler.c |  3 +++
 src/mesa/drivers/dri/i965/brw_nir.c      | 22 +++++++++++++++++-----
 2 files changed, 20 insertions(+), 5 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_compiler.c b/src/mesa/drivers/dri/i965/brw_compiler.c
index 6a73719..d7900a7 100644
--- a/src/mesa/drivers/dri/i965/brw_compiler.c
+++ b/src/mesa/drivers/dri/i965/brw_compiler.c
@@ -55,6 +55,7 @@ static const struct nir_shader_compiler_options scalar_nir_options = {
    .lower_unpack_snorm_4x8 = true,
    .lower_unpack_unorm_2x16 = true,
    .lower_unpack_unorm_4x8 = true,
+   .max_unroll_iterations = 32,
 };
 
 static const struct nir_shader_compiler_options vector_nir_options = {
@@ -75,6 +76,7 @@ static const struct nir_shader_compiler_options vector_nir_options = {
    .lower_unpack_unorm_2x16 = true,
    .lower_extract_byte = true,
    .lower_extract_word = true,
+   .max_unroll_iterations = 32,
 };
 
 static const struct nir_shader_compiler_options vector_nir_options_gen6 = {
@@ -92,6 +94,7 @@ static const struct nir_shader_compiler_options vector_nir_options_gen6 = {
    .lower_unpack_unorm_2x16 = true,
    .lower_extract_byte = true,
    .lower_extract_word = true,
+   .max_unroll_iterations = 32,
 };
 
 struct brw_compiler *
diff --git a/src/mesa/drivers/dri/i965/brw_nir.c b/src/mesa/drivers/dri/i965/brw_nir.c
index 9a217ad..a52d614 100644
--- a/src/mesa/drivers/dri/i965/brw_nir.c
+++ b/src/mesa/drivers/dri/i965/brw_nir.c
@@ -396,8 +396,17 @@ brw_nir_lower_cs_shared(nir_shader *nir)
 #define OPT_V(pass, ...) NIR_PASS_V(nir, pass, ##__VA_ARGS__)
 
 static nir_shader *
-nir_optimize(nir_shader *nir, bool is_scalar)
+nir_optimize(nir_shader *nir, const struct brw_compiler *compiler,
+             bool is_scalar)
 {
+   nir_variable_mode indirect_mask = 0;
+   if (compiler->glsl_compiler_options[nir->stage].EmitNoIndirectInput)
+      indirect_mask |= nir_var_shader_in;
+   if (compiler->glsl_compiler_options[nir->stage].EmitNoIndirectOutput)
+      indirect_mask |= nir_var_shader_out;
+   if (compiler->glsl_compiler_options[nir->stage].EmitNoIndirectTemp)
+      indirect_mask |= nir_var_local;
+
    bool progress;
    do {
       progress = false;
@@ -420,6 +429,9 @@ nir_optimize(nir_shader *nir, bool is_scalar)
       OPT(nir_opt_algebraic);
       OPT(nir_opt_constant_folding);
       OPT(nir_opt_dead_cf);
+      if (nir->options->max_unroll_iterations != 0) {
+         OPT(nir_opt_loop_unroll, indirect_mask);
+      }
       OPT(nir_opt_remove_phis);
       OPT(nir_opt_undef);
       OPT_V(nir_lower_doubles, nir_lower_drcp |
@@ -477,7 +489,7 @@ brw_preprocess_nir(const struct brw_compiler *compiler, nir_shader *nir)
 
    OPT(nir_split_var_copies);
 
-   nir = nir_optimize(nir, is_scalar);
+   nir = nir_optimize(nir, compiler, is_scalar);
 
    if (is_scalar) {
       OPT_V(nir_lower_load_const_to_scalar);
@@ -499,7 +511,7 @@ brw_preprocess_nir(const struct brw_compiler *compiler, nir_shader *nir)
    nir_lower_indirect_derefs(nir, indirect_mask);
 
    /* Get rid of split copies */
-   nir = nir_optimize(nir, is_scalar);
+   nir = nir_optimize(nir, compiler, is_scalar);
 
    OPT(nir_remove_dead_variables, nir_var_local);
 
@@ -524,7 +536,7 @@ brw_postprocess_nir(nir_shader *nir, const struct brw_compiler *compiler,
    bool progress; /* Written by OPT and OPT_V */
    (void)progress;
 
-   nir = nir_optimize(nir, is_scalar);
+   nir = nir_optimize(nir, compiler, is_scalar);
 
    if (devinfo->gen >= 6) {
       /* Try and fuse multiply-adds */
@@ -616,7 +628,7 @@ brw_nir_apply_sampler_key(nir_shader *nir,
 
    if (nir_lower_tex(nir, &tex_options)) {
       nir_validate_shader(nir);
-      nir = nir_optimize(nir, is_scalar);
+      nir = nir_optimize(nir, compiler, is_scalar);
    }
 
    return nir;
-- 
2.7.4