diff --git a/src/compiler/nir/nir_divergence_analysis.c b/src/compiler/nir/nir_divergence_analysis.c
index 0145b9073fe..3a170597ff9 100644
--- a/src/compiler/nir/nir_divergence_analysis.c
+++ b/src/compiler/nir/nir_divergence_analysis.c
@@ -542,6 +542,9 @@ visit_intrinsic(nir_intrinsic_instr *instr, struct divergence_state *state)
    case nir_intrinsic_shuffle_xor:
    case nir_intrinsic_shuffle_up:
    case nir_intrinsic_shuffle_down:
+   case nir_intrinsic_shuffle_xor_uniform_ir3:
+   case nir_intrinsic_shuffle_up_uniform_ir3:
+   case nir_intrinsic_shuffle_down_uniform_ir3:
    case nir_intrinsic_quad_broadcast:
    case nir_intrinsic_quad_swap_horizontal:
    case nir_intrinsic_quad_swap_vertical:
diff --git a/src/compiler/nir/nir_intrinsics.py b/src/compiler/nir/nir_intrinsics.py
index e756ef2b880..de24a64927d 100644
--- a/src/compiler/nir/nir_intrinsics.py
+++ b/src/compiler/nir/nir_intrinsics.py
@@ -1436,6 +1436,15 @@ intrinsic("inclusive_scan_clusters_ir3", dest_comp=1, src_comp=[1],
 intrinsic("exclusive_scan_clusters_ir3", dest_comp=1, src_comp=[1, 1],
           bit_sizes=src0, indices=[REDUCTION_OP])
 
+# Like shuffle_{xor,up,down} except with a uniform index. Necessary since the
+# ir3 shfl instruction doesn't work with divergent indices.
+intrinsic("shuffle_xor_uniform_ir3", src_comp=[0, 1], dest_comp=0,
+          bit_sizes=src0, flags=[CAN_ELIMINATE])
+intrinsic("shuffle_up_uniform_ir3", src_comp=[0, 1], dest_comp=0,
+          bit_sizes=src0, flags=[CAN_ELIMINATE])
+intrinsic("shuffle_down_uniform_ir3", src_comp=[0, 1], dest_comp=0,
+          bit_sizes=src0, flags=[CAN_ELIMINATE])
+
 # IR3-specific intrinsics for prefetching descriptors in preambles.
 intrinsic("prefetch_sam_ir3", [1, 1], flags=[CAN_REORDER])
 intrinsic("prefetch_tex_ir3", [1], flags=[CAN_REORDER])
diff --git a/src/compiler/nir/nir_lower_int64.c b/src/compiler/nir/nir_lower_int64.c
index 1d18210f321..6782385b989 100644
--- a/src/compiler/nir/nir_lower_int64.c
+++ b/src/compiler/nir/nir_lower_int64.c
@@ -1261,6 +1261,9 @@ should_lower_int64_intrinsic(const nir_intrinsic_instr *intrin,
    case nir_intrinsic_shuffle_xor:
    case nir_intrinsic_shuffle_up:
    case nir_intrinsic_shuffle_down:
+   case nir_intrinsic_shuffle_xor_uniform_ir3:
+   case nir_intrinsic_shuffle_up_uniform_ir3:
+   case nir_intrinsic_shuffle_down_uniform_ir3:
    case nir_intrinsic_rotate:
    case nir_intrinsic_quad_broadcast:
    case nir_intrinsic_quad_swap_horizontal:
@@ -1307,6 +1310,9 @@ lower_int64_intrinsic(nir_builder *b, nir_intrinsic_instr *intrin)
    case nir_intrinsic_shuffle_xor:
    case nir_intrinsic_shuffle_up:
    case nir_intrinsic_shuffle_down:
+   case nir_intrinsic_shuffle_xor_uniform_ir3:
+   case nir_intrinsic_shuffle_up_uniform_ir3:
+   case nir_intrinsic_shuffle_down_uniform_ir3:
    case nir_intrinsic_rotate:
    case nir_intrinsic_quad_broadcast:
    case nir_intrinsic_quad_swap_horizontal: