diff --git a/llvm/lib/Target/AMDGPU/VOP3PInstructions.td b/llvm/lib/Target/AMDGPU/VOP3PInstructions.td
index 24d7550e2dec4..8e42a4a454a70 100644
--- a/llvm/lib/Target/AMDGPU/VOP3PInstructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3PInstructions.td
@@ -360,7 +360,7 @@ class UDot2Pat<Instruction Inst> : GCNPat <
                              (and i32:$src1, (i32 65535)))
    ),
   (Inst (i32 8), $src0, (i32 8), $src1, (i32 8), $src2, (i1 0))> {
-  let SubtargetPredicate = !cast<VOP_Pseudo>(Inst).SubtargetPredicate;
+  let Predicates = !cast<VOP_Pseudo>(Inst).Predicates;
 }
 
 class SDot2Pat<Instruction Inst> : GCNPat <
@@ -369,40 +369,35 @@ class SDot2Pat<Instruction Inst> : GCNPat <
        (AMDGPUmul_i24_oneuse (sext_inreg i32:$src0, i16),
                              (sext_inreg i32:$src1, i16))),
   (Inst (i32 8), $src0, (i32 8), $src1, (i32 8), $src2, (i1 0))> {
-  let SubtargetPredicate = !cast<VOP_Pseudo>(Inst).SubtargetPredicate;
+  let Predicates = !cast<VOP_Pseudo>(Inst).Predicates;
 }
 
 let IsDOT = 1 in {
-let SubtargetPredicate = HasDot2Insts in {
-
+let OtherPredicates = [HasDot2Insts] in {
 defm V_DOT2_I32_I16 : VOP3PInst<"v_dot2_i32_i16",
   VOP3P_Profile<VOP_I32_V2I16_V2I16_I32>, int_amdgcn_sdot2, 1>;
 defm V_DOT2_U32_U16 : VOP3PInst<"v_dot2_u32_u16",
   VOP3P_Profile<VOP_I32_V2I16_V2I16_I32>, int_amdgcn_udot2, 1>;
+} // End OtherPredicates = [HasDot2Insts]
 
-} // End SubtargetPredicate = HasDot2Insts
-
-let SubtargetPredicate = HasDot10Insts in
+let OtherPredicates = [HasDot10Insts] in
 defm V_DOT2_F32_F16 : VOP3PInst<"v_dot2_f32_f16",
   VOP3P_Profile<VOP_F32_V2F16_V2F16_F32, VOP3_REGULAR, /*HasDPP*/ 1>,
   AMDGPUfdot2, 1/*ExplicitClamp*/>;
 
-let SubtargetPredicate = HasDot7Insts in {
+let OtherPredicates = [HasDot7Insts] in {
 defm V_DOT4_U32_U8  : VOP3PInst<"v_dot4_u32_u8",
   VOP3P_Profile<VOP_I32_I32_I32_I32, VOP3_PACKED>, int_amdgcn_udot4, 1>;
 defm V_DOT8_U32_U4  : VOP3PInst<"v_dot8_u32_u4",
   VOP3P_Profile<VOP_I32_I32_I32_I32, VOP3_PACKED>, int_amdgcn_udot8, 1>;
+} // End OtherPredicates = [HasDot7Insts]
 
-} // End SubtargetPredicate = HasDot7Insts
-
-let SubtargetPredicate = HasDot1Insts in {
-
+let OtherPredicates = [HasDot1Insts] in {
 defm V_DOT4_I32_I8  : VOP3PInst<"v_dot4_i32_i8",
   VOP3P_Profile<VOP_I32_I32_I32_I32, VOP3_PACKED>, int_amdgcn_sdot4, 1>;
 defm V_DOT8_I32_I4  : VOP3PInst<"v_dot8_i32_i4",
   VOP3P_Profile<VOP_I32_I32_I32_I32, VOP3_PACKED>, int_amdgcn_sdot8, 1>;
-
-} // End SubtargetPredicate = HasDot1Insts
+} // End OtherPredicates = [HasDot1Insts]
 
 def DOT2_BF16_Profile
   : VOP3P_Profile<VOP_F32_V2I16_V2I16_F32, VOP3_REGULAR, /*HasDPP*/ 1> {
@@ -456,14 +451,14 @@ def : UDot2Pat<V_DOT2_U32_U16>;
 def : SDot2Pat<V_DOT2_I32_I16>;
 
 foreach Type = ["U", "I"] in
-  let SubtargetPredicate = !cast<VOP_Pseudo>("V_DOT4_"#Type#"32_"#Type#8).SubtargetPredicate in
+  let Predicates = !cast<VOP_Pseudo>("V_DOT4_"#Type#"32_"#Type#8).Predicates in
   def : GCNPat <
     !cast<dag>(!foldl((i32 i32:$src2), [0, 1, 2, 3], lhs, y,
                       (add_oneuse lhs, (!cast<PatFrag>("Mul"#Type#"_Elt"#y) i32:$src0, i32:$src1)))),
     (!cast<VOP3P_Pseudo>("V_DOT4_"#Type#"32_"#Type#8) (i32 8), $src0, (i32 8), $src1, (i32 8), $src2, (i1 0))>;
 
 foreach Type = ["U", "I"] in
-  let SubtargetPredicate = !cast<VOP_Pseudo>("V_DOT8_"#Type#"32_"#Type#4).SubtargetPredicate in
+  let Predicates = !cast<VOP_Pseudo>("V_DOT8_"#Type#"32_"#Type#4).Predicates in
   def : GCNPat <
     !cast<dag>(!foldl((add_oneuse i32:$src2, (!cast<PatFrag>("Mul"#Type#"0_4bit") i32:$src0, i32:$src1)),
                       [1, 2, 3, 4, 5, 6, 7], lhs, y,
@@ -473,7 +468,7 @@ foreach Type = ["U", "I"] in
 // Different variants of dot8 code-gen dag patterns are not generated through table-gen due to a huge increase
 // in the compile time. Directly handle the pattern generated by the FE here.
 foreach Type = ["U", "I"] in
-  let SubtargetPredicate = !cast<VOP_Pseudo>("V_DOT8_"#Type#"32_"#Type#4).SubtargetPredicate in
+  let Predicates = !cast<VOP_Pseudo>("V_DOT8_"#Type#"32_"#Type#4).Predicates in
   def : GCNPat <
     !cast<dag>(!foldl((add_oneuse i32:$src2, (!cast<PatFrag>("Mul"#Type#"0_4bit") i32:$src0, i32:$src1)),
                       [7, 1, 2, 3, 4, 5, 6], lhs, y,
@@ -1128,16 +1123,15 @@ defm V_PK_ADD_F16 : VOP3P_Real_vi <0x0f>;
 defm V_PK_MUL_F16 : VOP3P_Real_vi <0x10>;
 defm V_PK_MIN_F16 : VOP3P_Real_vi <0x11>;
 defm V_PK_MAX_F16 : VOP3P_Real_vi <0x12>;
-} // End SubtargetPredicate = isGFX8GFX9
 
-let SubtargetPredicate = HasMadMixInsts in {
+let OtherPredicates = [HasMadMixInsts] in {
 defm V_MAD_MIX_F32 : VOP3P_Real_vi <0x20>;
 defm V_MAD_MIXLO_F16 : VOP3P_Real_vi <0x21>;
 defm V_MAD_MIXHI_F16 : VOP3P_Real_vi <0x22>;
 }
 
-let SubtargetPredicate = HasFmaMixInsts in {
-let DecoderNamespace = "GFX9_DL" in {
+let OtherPredicates = [HasFmaMixInsts],
+    DecoderNamespace = "GFX9_DL" in {
 // The mad_mix instructions were renamed and their behaviors changed,
 // but the opcode stayed the same so we need to put these in a
 // different DecoderNamespace to avoid the ambiguity.
@@ -1145,8 +1139,6 @@ defm V_FMA_MIX_F32 : VOP3P_Real_vi <0x20>;
 defm V_FMA_MIXLO_F16 : VOP3P_Real_vi <0x21>;
 defm V_FMA_MIXHI_F16 : VOP3P_Real_vi <0x22>;
 }
-}
-
 
 defm V_DOT2_I32_I16 : VOP3P_Real_vi <0x26>;
 defm V_DOT2_U32_U16 : VOP3P_Real_vi <0x27>;
@@ -1157,6 +1149,7 @@ defm V_DOT8_U32_U4  : VOP3P_Real_vi <0x2b>;
 
 defm V_DOT4_I32_I8  : VOP3P_Real_vi <0x28>;
 defm V_DOT8_I32_I4  : VOP3P_Real_vi <0x2a>;
+} // End SubtargetPredicate = isGFX8GFX9
 
 let OtherPredicates = [HasMAIInsts] in {