radeonsi/gfx10: simplify a streamout loop in gfx10_emit_ngg_epilogue

[mesa.git] / src / gallium / drivers / vc4 / vc4_program.c

diff --git a/src/gallium/drivers/vc4/vc4_program.c b/src/gallium/drivers/vc4/vc4_program.c
index 2ca3f907135ad9834832212ffda9ddc83f187be7..7fb25ba782e4b4bba9c36a0b28d818a82d3f1d9b 100644 (file)
--- a/src/gallium/drivers/vc4/vc4_program.c
+++ b/src/gallium/drivers/vc4/vc4_program.c
@@ -1128,8 +1128,7 @@ ntq_emit_alu(struct vc4_compile *c, nir_alu_instr *instr)
         struct qreg result;
 
         switch (instr->op) {
-        case nir_op_fmov:
-        case nir_op_imov:
+        case nir_op_mov:
                 result = qir_MOV(c, src[0]);
                 break;
         case nir_op_fmul:
@@ -1527,12 +1526,16 @@ static void
 vc4_optimize_nir(struct nir_shader *s)
 {
         bool progress;
+        unsigned lower_flrp =
+                (s->options->lower_flrp16 ? 16 : 0) |
+                (s->options->lower_flrp32 ? 32 : 0) |
+                (s->options->lower_flrp64 ? 64 : 0);
 
         do {
                 progress = false;
 
                 NIR_PASS_V(s, nir_lower_vars_to_ssa);
-                NIR_PASS(progress, s, nir_lower_alu_to_scalar);
+                NIR_PASS(progress, s, nir_lower_alu_to_scalar, NULL);
                 NIR_PASS(progress, s, nir_lower_phis_to_scalar);
                 NIR_PASS(progress, s, nir_copy_prop);
                 NIR_PASS(progress, s, nir_opt_remove_phis);
@@ -1542,6 +1545,24 @@ vc4_optimize_nir(struct nir_shader *s)
                 NIR_PASS(progress, s, nir_opt_peephole_select, 8, true, true);
                 NIR_PASS(progress, s, nir_opt_algebraic);
                 NIR_PASS(progress, s, nir_opt_constant_folding);
+                if (lower_flrp != 0) {
+                        bool lower_flrp_progress = false;
+
+                        NIR_PASS(lower_flrp_progress, s, nir_lower_flrp,
+                                 lower_flrp,
+                                 false /* always_precise */,
+                                 s->options->lower_ffma);
+                        if (lower_flrp_progress) {
+                                NIR_PASS(progress, s, nir_opt_constant_folding);
+                                progress = true;
+                        }
+
+                        /* Nothing should rematerialize any flrps, so we only
+                         * need to do this lowering once.
+                         */
+                        lower_flrp = 0;
+                }
+
                 NIR_PASS(progress, s, nir_opt_undef);
                 NIR_PASS(progress, s, nir_opt_loop_unroll,
                          nir_var_shader_in |
@@ -2168,11 +2189,13 @@ static const nir_shader_compiler_options nir_options = {
         .lower_fdiv = true,
         .lower_ffma = true,
         .lower_flrp32 = true,
+        .lower_fmod = true,
         .lower_fpow = true,
         .lower_fsat = true,
         .lower_fsqrt = true,
         .lower_ldexp = true,
         .lower_negate = true,
+        .lower_rotate = true,
         .max_unroll_iterations = 32,
 };