diff --git a/src/broadcom/compiler/nir_to_vir.c b/src/broadcom/compiler/nir_to_vir.c
index 0303ca96103..ec6427f3d47 100644
--- a/src/broadcom/compiler/nir_to_vir.c
+++ b/src/broadcom/compiler/nir_to_vir.c
@@ -656,9 +656,10 @@ ntq_emit_tmu_general(struct v3d_compile *c, nir_intrinsic_instr *instr,
                          */
                         uint32_t perquad =
                                 is_load && !vir_in_nonuniform_control_flow(c) &&
-                                c->s->info.stage == MESA_SHADER_FRAGMENT &&
-                                c->s->info.fs.needs_quad_helper_invocations &&
-                                !c->emitted_discard ?
+                                ((c->s->info.stage == MESA_SHADER_FRAGMENT &&
+                                  c->s->info.fs.needs_quad_helper_invocations &&
+                                  !c->emitted_discard) ||
+                                 c->s->info.uses_wide_subgroup_intrinsics) ?
                                 GENERAL_TMU_LOOKUP_PER_QUAD :
                                 GENERAL_TMU_LOOKUP_PER_PIXEL;
                         config = 0xffffff00 | tmu_op << 3 | perquad;