diff --git a/src/backend/x64/emit_x64_vector.cpp b/src/backend/x64/emit_x64_vector.cpp
index 7e7f7160..186170ec 100644
--- a/src/backend/x64/emit_x64_vector.cpp
+++ b/src/backend/x64/emit_x64_vector.cpp
@@ -1465,20 +1465,21 @@ void EmitX64::EmitVectorLogicalShiftLeft8(EmitContext& ctx, IR::Inst* inst) {
     const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
     const u8 shift_amount = args[1].GetImmediateU8();
 
-    if (shift_amount == 1) {
+    if (shift_amount == 0) {
+        // do nothing
+    } else if (shift_amount >= 8) {
+        code.pxor(result, result);
+    } else if (shift_amount == 1) {
         code.paddb(result, result);
-    } else if (shift_amount > 0) {
-        if (code.HasAVX512_Icelake()) {
-            // Galois 8x8 identity matrix, bit-shifted by the shift-amount
-            const u64 shift_matrix = 0x0102040810204080 >> (shift_amount * 8);
-            code.vgf2p8affineqb(result, result, code.MConst(xword_b, shift_matrix), 0);
-        } else {
-            const u64 replicand = (0xFFULL << shift_amount) & 0xFF;
-            const u64 mask = Common::Replicate(replicand, Common::BitSize<u8>());
+    } else if (code.HasAVX512_Icelake()) {
+        const u64 shift_matrix = 0x0102040810204080 >> (shift_amount * 8);
+        code.vgf2p8affineqb(result, result, code.MConst(xword_b, shift_matrix), 0);
+    } else {
+        const u64 replicand = (0xFFULL << shift_amount) & 0xFF;
+        const u64 mask = Common::Replicate(replicand, Common::BitSize<u8>());
 
-            code.psllw(result, shift_amount);
-            code.pand(result, code.MConst(xword, mask, mask));
-        }
+        code.psllw(result, shift_amount);
+        code.pand(result, code.MConst(xword, mask, mask));
     }
 
     ctx.reg_alloc.DefineValue(inst, result);