diff --git a/src/nouveau/compiler/nak/hw_tests.rs b/src/nouveau/compiler/nak/hw_tests.rs
index f5790361265..0b7853faa0e 100644
--- a/src/nouveau/compiler/nak/hw_tests.rs
+++ b/src/nouveau/compiler/nak/hw_tests.rs
@@ -566,6 +566,40 @@ fn test_op_iadd3x() {
     }
 }
 
+#[test]
+fn test_op_shf() {
+    let sm = &RunSingleton::get().sm;
+
+    let types = [IntType::U32, IntType::I32, IntType::U64, IntType::I64];
+
+    for i in 0..32 {
+        let op = OpShf {
+            dst: Dst::None,
+            low: 0.into(),
+            high: 0.into(),
+            shift: 0.into(),
+            data_type: types[i & 0x3],
+            right: i & 0x4 != 0,
+            wrap: i & 0x8 != 0,
+            dst_high: i & 0x10 != 0,
+        };
+
+        if sm.sm() < 70 && !(op.dst_high || op.right) {
+            continue;
+        }
+
+        let shift_idx = op.src_idx(&op.shift);
+        let mut a = Acorn::new();
+        test_foldable_op_with(op, &mut |i| {
+            if i == shift_idx {
+                a.get_uint(6) as u32
+            } else {
+                a.get_u32()
+            }
+        });
+    }
+}
+
 #[test]
 fn test_op_prmt() {
     let op = OpPrmt {
diff --git a/src/nouveau/compiler/nak/ir.rs b/src/nouveau/compiler/nak/ir.rs
index d7729538f76..7e00da7106a 100644
--- a/src/nouveau/compiler/nak/ir.rs
+++ b/src/nouveau/compiler/nak/ir.rs
@@ -2071,7 +2071,7 @@ impl fmt::Display for ImageDim {
     }
 }
 
-#[derive(Clone, Copy, Eq, Hash, PartialEq)]
+#[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]
 pub enum IntType {
     U8,
     I8,
@@ -3616,7 +3616,7 @@ impl fmt::Display for ShflOp {
 }
 
 #[repr(C)]
-#[derive(SrcsAsSlice, DstsAsSlice)]
+#[derive(Clone, SrcsAsSlice, DstsAsSlice)]
 pub struct OpShf {
     #[dst_type(GPR)]
     pub dst: Dst,
@@ -3636,6 +3636,55 @@ pub struct OpShf {
     pub dst_high: bool,
 }
 
+impl Foldable for OpShf {
+    fn fold(&self, sm: &dyn ShaderModel, f: &mut OpFoldData<'_>) {
+        let low = f.get_u32_src(self, &self.low);
+        let high = f.get_u32_src(self, &self.high);
+        let shift = f.get_u32_src(self, &self.shift);
+
+        let bits: u32 = self.data_type.bits().try_into().unwrap();
+        let shift = if self.wrap {
+            shift & (bits - 1)
+        } else {
+            min(shift, bits)
+        };
+
+        let x = u64::from(low) | (u64::from(high) << 32);
+        let shifted = if sm.sm() < 70
+            && self.dst_high
+            && self.data_type != IntType::I64
+        {
+            if self.right {
+                x.checked_shr(shift).unwrap_or(0) as u64
+            } else {
+                x.checked_shl(shift).unwrap_or(0) as u64
+            }
+        } else if self.data_type.is_signed() {
+            if self.right {
+                (x as i64).checked_shr(shift).unwrap_or(0) as u64
+            } else {
+                (x as i64).checked_shl(shift).unwrap_or(0) as u64
+            }
+        } else {
+            if self.right {
+                x.checked_shr(shift).unwrap_or(0) as u64
+            } else {
+                x.checked_shl(shift).unwrap_or(0) as u64
+            }
+        };
+
+        let dst = if sm.sm() < 70 && !self.right {
+            (shifted >> 32) as u32
+        } else if self.dst_high {
+            (shifted >> 32) as u32
+        } else {
+            shifted as u32
+        };
+
+        f.set_u32_dst(self, &self.dst, dst);
+    }
+}
+
 impl DisplayOp for OpShf {
     fn fmt_op(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         write!(f, "shf")?;