i965/fs: Allow SIMD16 on pre-SNB when try_replace_with_sel is successful

If try_replace_with_sel is able to replace the flow control with a SEL
instruction, then there is no flow control... failing SIMD16 because
of nonexistent flow control is wrong.

No piglit regressions on any i965 platform in Jenkins.

total instructions in shared programs: 4382707 -> 4382707 (0.00%)
instructions in affected programs:     0 -> 0
helped:                                0
HURT:                                  0
GAINED:                                2089
LOST:                                  0

No other platforms affected in shader-db.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h
index 1de10bb..419fe48 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs.h
+++ b/src/mesa/drivers/dri/i965/brw_fs.h
@@ -535,7 +535,7 @@ public:
    bool try_emit_saturate(ir_expression *ir);
    bool try_emit_line(ir_expression *ir);
    bool try_emit_mad(ir_expression *ir);
-   void try_replace_with_sel();
+   bool try_replace_with_sel();
    bool opt_peephole_sel();
    bool opt_peephole_predicated_break();
    bool opt_saturate_propagation();

diff --git a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
index 0eb137f..de0d780 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
@@ -435,10 +435,6 @@ fs_visitor::nir_emit_cf_list(exec_list *list)
 void
 fs_visitor::nir_emit_if(nir_if *if_stmt)
 {
-   if (brw->gen < 6) {
-      no16("Can't support (non-uniform) control flow on SIMD16\n");
-   }
-
    /* first, put the condition into f0 */
    fs_inst *inst = emit(MOV(reg_null_d,
                             retype(get_nir_src(if_stmt->condition),
@@ -456,7 +452,9 @@ fs_visitor::nir_emit_if(nir_if *if_stmt)
 
    emit(BRW_OPCODE_ENDIF);
 
-   try_replace_with_sel();
+   if (!try_replace_with_sel() && brw->gen < 6) {
+      no16("Can't support (non-uniform) control flow on SIMD16\n");
+   }
 }
 
 void

diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index 9805b55..f5d7383 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -2741,7 +2741,7 @@ fs_visitor::emit_if_gen6(ir_if *ir)
  *
  * If src0 is an immediate value, we promote it to a temporary GRF.
  */
-void
+bool
 fs_visitor::try_replace_with_sel()
 {
    fs_inst *endif_inst = (fs_inst *) instructions.get_tail();
@@ -2755,7 +2755,7 @@ fs_visitor::try_replace_with_sel()
    fs_inst *match = (fs_inst *) endif_inst->prev;
    for (int i = 0; i < 4; i++) {
       if (match->is_head_sentinel() || match->opcode != opcodes[4-i-1])
-         return;
+         return false;
       match = (fs_inst *) match->prev;
    }
 
@@ -2797,16 +2797,16 @@ fs_visitor::try_replace_with_sel()
          sel->predicate = if_inst->predicate;
          sel->predicate_inverse = if_inst->predicate_inverse;
       }
+
+      return true;
    }
+
+   return false;
 }
 
 void
 fs_visitor::visit(ir_if *ir)
 {
-   if (brw->gen < 6) {
-      no16("Can't support (non-uniform) control flow on SIMD16\n");
-   }
-
    /* Don't point the annotation at the if statement, because then it plus
     * the then and else blocks get printed.
     */
@@ -2836,7 +2836,9 @@ fs_visitor::visit(ir_if *ir)
 
    emit(BRW_OPCODE_ENDIF);
 
-   try_replace_with_sel();
+   if (!try_replace_with_sel() && brw->gen < 6) {
+      no16("Can't support (non-uniform) control flow on SIMD16\n");
+   }
 }
 
 void