diff --git a/src/compiler/nir/nir_divergence_analysis.c b/src/compiler/nir/nir_divergence_analysis.c
index 88f36250c10..6a7aef2701c 100644
--- a/src/compiler/nir/nir_divergence_analysis.c
+++ b/src/compiler/nir/nir_divergence_analysis.c
@@ -774,7 +774,6 @@ visit_intrinsic(nir_intrinsic_instr *instr, struct divergence_state *state)
    case nir_intrinsic_load_color0:
    case nir_intrinsic_load_color1:
    case nir_intrinsic_load_sample_id:
-   case nir_intrinsic_load_sample_id_no_per_sample:
    case nir_intrinsic_load_sample_mask_in:
    case nir_intrinsic_load_interpolated_input:
    case nir_intrinsic_load_point_coord_maybe_flipped:
diff --git a/src/compiler/nir/nir_intrinsics.py b/src/compiler/nir/nir_intrinsics.py
index d3db475e7b1..b8c8c16b6e1 100644
--- a/src/compiler/nir/nir_intrinsics.py
+++ b/src/compiler/nir/nir_intrinsics.py
@@ -925,9 +925,6 @@ system_value("instance_id", 1)
 system_value("base_instance", 1)
 system_value("draw_id", 1)
 system_value("sample_id", 1)
-# sample_id_no_per_sample is like sample_id but does not imply per-
-# sample shading.  See the lower_helper_invocation option.
-system_value("sample_id_no_per_sample", 1)
 system_value("sample_pos", 2)
 # sample_pos_or_center is like sample_pos but does not imply per-sample
 # shading.  When per-sample dispatch is not enabled, it returns (0.5, 0.5).
diff --git a/src/freedreno/ir3/ir3_compiler_nir.c b/src/freedreno/ir3/ir3_compiler_nir.c
index ecc9ab536f8..acfca72a33d 100644
--- a/src/freedreno/ir3/ir3_compiler_nir.c
+++ b/src/freedreno/ir3/ir3_compiler_nir.c
@@ -2968,7 +2968,6 @@ emit_intrinsic(struct ir3_context *ctx, nir_intrinsic_instr *intr)
       dst[0] = ctx->instance_id;
       break;
    case nir_intrinsic_load_sample_id:
-   case nir_intrinsic_load_sample_id_no_per_sample:
       if (!ctx->samp_id) {
          ctx->samp_id = create_sysval_input(ctx, SYSTEM_VALUE_SAMPLE_ID, 0x1);
          ctx->samp_id->dsts[0]->flags |= IR3_REG_HALF;
diff --git a/src/microsoft/compiler/nir_to_dxil.c b/src/microsoft/compiler/nir_to_dxil.c
index b62f8506fee..0e7b5f91ae2 100644
--- a/src/microsoft/compiler/nir_to_dxil.c
+++ b/src/microsoft/compiler/nir_to_dxil.c
@@ -4562,15 +4562,10 @@ emit_load_sample_pos_from_id(struct ntd_context *ctx, nir_intrinsic_instr *intr)
 static bool
 emit_load_sample_id(struct ntd_context *ctx, nir_intrinsic_instr *intr)
 {
-   assert(ctx->mod.info.has_per_sample_input ||
-          intr->intrinsic == nir_intrinsic_load_sample_id_no_per_sample);
+   assert(ctx->mod.info.has_per_sample_input);
 
-   if (ctx->mod.info.has_per_sample_input)
-      return emit_load_unary_external_function(ctx, intr, "dx.op.sampleIndex",
-                                               DXIL_INTR_SAMPLE_INDEX, nir_type_int);
-
-   store_def(ctx, &intr->def, 0, dxil_module_get_int32_const(&ctx->mod, 0));
-   return true;
+   return emit_load_unary_external_function(ctx, intr, "dx.op.sampleIndex",
+                                            DXIL_INTR_SAMPLE_INDEX, nir_type_int);
 }
 
 static bool
@@ -4821,7 +4816,6 @@ emit_intrinsic(struct ntd_context *ctx, nir_intrinsic_instr *intr)
       return emit_load_unary_external_function(ctx, intr, "dx.op.primitiveID",
                                                DXIL_INTR_PRIMITIVE_ID, nir_type_int);
    case nir_intrinsic_load_sample_id:
-   case nir_intrinsic_load_sample_id_no_per_sample:
       return emit_load_sample_id(ctx, intr);
    case nir_intrinsic_load_invocation_id:
       switch (ctx->mod.shader_kind) {
diff --git a/src/panfrost/util/pan_lower_helper_invocation.c b/src/panfrost/util/pan_lower_helper_invocation.c
index 4c75ae948fe..80035788dfb 100644
--- a/src/panfrost/util/pan_lower_helper_invocation.c
+++ b/src/panfrost/util/pan_lower_helper_invocation.c
@@ -25,10 +25,8 @@
 #include "pan_ir.h"
 
 /* Lower gl_HelperInvocation to (gl_SampleMaskIn == 0), this depends on
- * architectural details but is required for correct operation with
- * multisampling. NIR's lowering won't work for us, since there is no in-spec
- * way to implement load_sample_id_no_per_sample. */
-
+ * architectural details but is more efficient than NIR's lowering.
+ */
 static bool
 pan_lower_helper_invocation_instr(nir_builder *b, nir_intrinsic_instr *intr,
                                   void *data)