feat(arch): add the amd64 fp16 packed imm8-control group

Assisted-by: GLM 5.3

asm/extension_amd64_test.go

@@ -55,6 +55,9 @@ func TestAmd64ExtensionRegistry(t *testing.T) {
 		{"VCVTUQQ2PH", 3},
 		{"VCVTPH2PD", 3},
 		{"VCVTPD2PH", 3},
+		{"VRNDSCALEPH", 3},
+		{"VREDUCEPH", 3},
+		{"VGETMANTPH", 3},
 	} {
 		cands, ok := LookupExtension(arch.AMD64, tt.mnem)
 		if !ok {
@@ -68,8 +71,8 @@ func TestAmd64ExtensionRegistry(t *testing.T) {
 			t.Errorf("the %s lookup is not case-insensitive", tt.mnem)
 		}
 	}
-	if got := arch.Extensions(arch.AMD64); len(got) != 112 {
-		t.Errorf("the amd64 layer registers %d instructions, want 112", len(got))
+	if got := arch.Extensions(arch.AMD64); len(got) != 121 {
+		t.Errorf("the amd64 layer registers %d instructions, want 121", len(got))
 	}
 	if _, ok := LookupExtension(arch.AMD64, "NOSUCHINSTR"); ok {
 		t.Error("a non-extended mnemonic resolved")
@@ -160,6 +163,9 @@ func TestEncodeExtensionAmd64(t *testing.T) {
 		{"fp16 to double-precision, widened", "VCVTPH2PD",
 			[]arch.ExtOperand{arch.ExtXmm(5), arch.ExtZmm(6)},
 			"62f57c485af5"},
+		{"packed fp16 rounding to fraction bits", "VRNDSCALEPH",
+			[]arch.ExtOperand{arch.ExtImmediate(0x7b), arch.ExtZmm(5), arch.ExtZmm(6)},
+			"62f37c4808f57b"},
 	} {
 		got, err := EncodeExtension(arch.AMD64, tt.mnem, tt.ops...)
 		if err != nil {