nir: add ARB_shader_ballot and ARB_shader_group_vote instructionsnir-divergence-v3

v2: make every instruction an intrinsic, add missing subgroup_size,
also add system value translation stuff

3 files changed, 61 insertions, 0 deletions

diff --git a/src/compiler/nir/nir.c b/src/compiler/nir/nir.c
index 491b908396..266ca6d1be 100644
--- a/src/compiler/nir/nir.c
+++ b/src/compiler/nir/nir.c
@@ -1864,6 +1864,20 @@ nir_intrinsic_op
 nir_intrinsic_from_system_value(gl_system_value val)
 {
    switch (val) {
+   case SYSTEM_VALUE_SUBGROUP_SIZE:
+      return nir_intrinsic_load_subgroup_size;
+   case SYSTEM_VALUE_SUBGROUP_INVOCATION:
+      return nir_intrinsic_load_subgroup_invocation;
+   case SYSTEM_VALUE_SUBGROUP_EQ_MASK:
+      return nir_intrinsic_load_subgroup_eq_mask;
+   case SYSTEM_VALUE_SUBGROUP_GE_MASK:
+      return nir_intrinsic_load_subgroup_ge_mask;
+   case SYSTEM_VALUE_SUBGROUP_GT_MASK:
+      return nir_intrinsic_load_subgroup_gt_mask;
+   case SYSTEM_VALUE_SUBGROUP_LE_MASK:
+      return nir_intrinsic_load_subgroup_le_mask;
+   case SYSTEM_VALUE_SUBGROUP_LT_MASK:
+      return nir_intrinsic_load_subgroup_lt_mask;
    case SYSTEM_VALUE_VERTEX_ID:
       return nir_intrinsic_load_vertex_id;
    case SYSTEM_VALUE_INSTANCE_ID:
@@ -1917,6 +1931,20 @@ gl_system_value
 nir_system_value_from_intrinsic(nir_intrinsic_op intrin)
 {
    switch (intrin) {
+   case nir_intrinsic_load_subgroup_size:
+      return SYSTEM_VALUE_SUBGROUP_SIZE;
+   case nir_intrinsic_load_subgroup_invocation:
+      return SYSTEM_VALUE_SUBGROUP_INVOCATION;
+   case nir_intrinsic_load_subgroup_eq_mask:
+      return SYSTEM_VALUE_SUBGROUP_EQ_MASK;
+   case nir_intrinsic_load_subgroup_ge_mask:
+      return SYSTEM_VALUE_SUBGROUP_GE_MASK;
+   case nir_intrinsic_load_subgroup_gt_mask:
+      return SYSTEM_VALUE_SUBGROUP_GT_MASK;
+   case nir_intrinsic_load_subgroup_le_mask:
+      return SYSTEM_VALUE_SUBGROUP_LE_MASK;
+   case nir_intrinsic_load_subgroup_lt_mask:
+      return SYSTEM_VALUE_SUBGROUP_LT_MASK;
    case nir_intrinsic_load_vertex_id:
       return SYSTEM_VALUE_VERTEX_ID;
    case nir_intrinsic_load_instance_id:
diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index 64caccb621..f87e1c27c9 100644
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -1920,6 +1920,9 @@ typedef struct nir_shader_compiler_options {
    bool use_interpolated_input_intrinsics;
 
    unsigned max_unroll_iterations;
+
+   /* Number of channels in a subgroup. */
+   unsigned subgroup_size;
 } nir_shader_compiler_options;
 
 typedef struct nir_shader {
diff --git a/src/compiler/nir/nir_intrinsics.h b/src/compiler/nir/nir_intrinsics.h
index 21e7d904b7..62e488428c 100644
--- a/src/compiler/nir/nir_intrinsics.h
+++ b/src/compiler/nir/nir_intrinsics.h
@@ -331,6 +331,36 @@ SYSTEM_VALUE(alpha_ref_float, 1, 0, xx, xx, xx)
 SYSTEM_VALUE(layer_id, 1, 0, xx, xx, xx)
 SYSTEM_VALUE(view_index, 1, 0, xx, xx, xx)
 
+
+/* ARB_shader_ballot instructions */
+
+SYSTEM_VALUE(subgroup_size, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_invocation, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_eq_mask, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_ge_mask, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_gt_mask, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_le_mask, 1, 0, xx, xx, xx)
+SYSTEM_VALUE(subgroup_lt_mask, 1, 0, xx, xx, xx)
+
+#define CROSS_THREAD(name, dest_components, src_components) \
+   INTRINSIC(name, 1, ARR(src_components), true, dest_components, 0, 0, \
+             xx, xx, xx, \
+             NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER | \
+             NIR_INTRINSIC_CROSS_THREAD)
+
+CROSS_THREAD(ballot, 0, 0)
+INTRINSIC(read_invocation, 2, ARR(0, 1), true, 0, 0, 0,
+          xx, xx, xx,
+          NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER |
+          NIR_INTRINSIC_CROSS_THREAD)
+CROSS_THREAD(read_first_invocation, 0, 0)
+
+/* ARB_shader_group_vote instructions */
+
+CROSS_THREAD(any_invocations, 1, 1)
+CROSS_THREAD(all_invocations, 1, 1)
+CROSS_THREAD(all_invocations_equal, 1, 1)
+
 /* Blend constant color values.  Float values are clamped. */
 SYSTEM_VALUE(blend_const_color_r_float, 1, 0, xx, xx, xx)
 SYSTEM_VALUE(blend_const_color_g_float, 1, 0, xx, xx, xx)