diff --git a/.pick_status.json b/.pick_status.json
index 406eb063dff..a267826a5f3 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -784,7 +784,7 @@
         "description": "intel/nir: Don't needlessly split u2f16 for nir_type_uint32",
         "nominated": true,
         "nomination_type": 1,
-        "resolution": 0,
+        "resolution": 1,
         "main_sha": null,
         "because_sha": "f695a9fed2b0eac39acdaca132f5fc8c43c9f7d7",
         "notes": null
diff --git a/src/intel/compiler/intel_nir_lower_conversions.c b/src/intel/compiler/intel_nir_lower_conversions.c
index e0dde853349..ba8af15babb 100644
--- a/src/intel/compiler/intel_nir_lower_conversions.c
+++ b/src/intel/compiler/intel_nir_lower_conversions.c
@@ -60,9 +60,8 @@ lower_alu_instr(nir_builder *b, nir_alu_instr *alu)
     * 32-bit float type so we don't lose range when we convert from
     * a 64-bit integer.
     */
-   unsigned int64_types = nir_type_int64 | nir_type_uint64;
-   if ((src_full_type == nir_type_float16 && (dst_full_type & int64_types)) ||
-       ((src_full_type & int64_types) && dst_full_type == nir_type_float16)) {
+   if ((src_full_type == nir_type_float16 && dst_bit_size == 64) ||
+       (src_bit_size == 64 && dst_full_type == nir_type_float16)) {
       split_conversion(b, alu, src_type, nir_type_float | 32,
                        dst_type | dst_bit_size);
       return true;