diff --git a/src/panfrost/bifrost/bifrost_compile.c b/src/panfrost/bifrost/bifrost_compile.c
index 52b5eecd90f..ca1d33f862b 100644
--- a/src/panfrost/bifrost/bifrost_compile.c
+++ b/src/panfrost/bifrost/bifrost_compile.c
@@ -1969,6 +1969,10 @@ bi_emit_alu(bi_builder *b, nir_alu_instr *instr)
                 bi_hadd_to(b, nir_type_int, sz, dst, s0, s1, BI_ROUND_RTP);
                 break;
 
+        case nir_op_ineg:
+                bi_isub_to(b, nir_type_int, sz, dst, bi_zero(), s0, false);
+                break;
+
         case nir_op_isub:
                 bi_isub_to(b, nir_type_int, sz, dst, s0, s1, false);
                 break;
diff --git a/src/panfrost/bifrost/bifrost_nir_algebraic.py b/src/panfrost/bifrost/bifrost_nir_algebraic.py
index 81d2708573a..5dd553cb685 100644
--- a/src/panfrost/bifrost/bifrost_nir_algebraic.py
+++ b/src/panfrost/bifrost/bifrost_nir_algebraic.py
@@ -30,13 +30,7 @@ a = 'a'
 b = 'b'
 c = 'c'
 
-algebraic_late = [
-    # ineg must be lowered late, but only for integers; floats will try to
-    # have modifiers attached... hence why this has to be here rather than
-    # a more standard lower_negate approach
-
-    (('ineg', a), ('isub', 0, a)),
-]
+algebraic_late = []
 
 for isz in ('8', '16', '32'):
         for osz in ('16', '32', '64'):