[AMDGPU] Supress unused sdwa insts generation

Do not generate non-existing sdwa instructions. It reduces the
number of generated instructions by 185.

Differential Revision: https://reviews.llvm.org/D69010

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375016 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AMDGPU/VOP1Instructions.td b/lib/Target/AMDGPU/VOP1Instructions.td
index 0659b1c2bdf68ab8e1a58ea84fd884ed27ccfeb9..f1cdc3097dc04c7558fbfea1e2600879977dff64 100644 (file)
--- a/lib/Target/AMDGPU/VOP1Instructions.td
+++ b/lib/Target/AMDGPU/VOP1Instructions.td
@@ -104,13 +104,18 @@ multiclass VOP1Inst <string opName, VOPProfile P,
                      SDPatternOperator node = null_frag> {
   def _e32 : VOP1_Pseudo <opName, P>;
   def _e64 : VOP3_Pseudo <opName, P, getVOP1Pat64<node, P>.ret>;
-  def _sdwa : VOP1_SDWA_Pseudo <opName, P>;
+
+  foreach _ = BoolToList<P.HasExtSDWA>.ret in
+    def _sdwa : VOP1_SDWA_Pseudo <opName, P>;
+
   foreach _ = BoolToList<P.HasExtDPP>.ret in
     def _dpp : VOP1_DPP_Pseudo <opName, P>;
 
   def : MnemonicAlias<opName#"_e32", opName>, LetDummies;
   def : MnemonicAlias<opName#"_e64", opName>, LetDummies;
-  def : MnemonicAlias<opName#"_sdwa", opName>, LetDummies;
+
+  foreach _ = BoolToList<P.HasExtSDWA>.ret in
+    def : MnemonicAlias<opName#"_sdwa", opName>, LetDummies;
 
   foreach _ = BoolToList<P.HasExtDPP>.ret in
     def : MnemonicAlias<opName#"_dpp", opName>, LetDummies;
@@ -500,6 +505,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
       VOP3e_gfx10<{0, 1, 1, op{6-0}}, !cast<VOP3_Pseudo>(NAME#"_e64").Pfl>;
   }
   multiclass VOP1_Real_sdwa_gfx10<bits<9> op> {
+    foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
     def _sdwa_gfx10 :
       VOP_SDWA10_Real<!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
       VOP1_SDWA9Ae<op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl> {
@@ -714,10 +720,12 @@ multiclass VOP1_Real_e32e64_vi <bits<10> op> {
 multiclass VOP1_Real_vi <bits<10> op> {
   defm NAME : VOP1_Real_e32e64_vi <op>;
 
+  foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA>.ret in
   def _sdwa_vi :
     VOP_SDWA_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOP1_SDWAe <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
 
+  foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
   def _sdwa_gfx9 :
     VOP_SDWA9_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOP1_SDWA9Ae <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
@@ -895,6 +903,7 @@ multiclass VOP1_Real_gfx9 <bits<10> op> {
     defm NAME : VOP1_Real_e32e64_vi <op>;
   }
 
+  foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
   def _sdwa_gfx9 :
     VOP_SDWA9_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOP1_SDWA9Ae <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;

diff --git a/lib/Target/AMDGPU/VOP2Instructions.td b/lib/Target/AMDGPU/VOP2Instructions.td
index c23755fe39a29c1e443db8cf3f3821c0941154c5..4b38a31a02d3431b5390450a158e23f075d03583 100644 (file)
--- a/lib/Target/AMDGPU/VOP2Instructions.td
+++ b/lib/Target/AMDGPU/VOP2Instructions.td
@@ -147,7 +147,8 @@ multiclass VOP2Inst_sdwa<string opName,
                          string revOp = opName,
                          bit GFX9Renamed = 0> {
   let renamedInGFX9 = GFX9Renamed in {
-    def _sdwa : VOP2_SDWA_Pseudo <opName, P>;
+    foreach _ = BoolToList<P.HasExtSDWA>.ret in
+      def _sdwa : VOP2_SDWA_Pseudo <opName, P>;
   } // End renamedInGFX9 = GFX9Renamed
 }
 
@@ -179,9 +180,10 @@ multiclass VOP2bInst <string opName,
           let usesCustomInserter = !eq(P.NumSrcArgs, 2);
         }
 
-        def _sdwa  : VOP2_SDWA_Pseudo <opName, P> {
-          let AsmMatchConverter = "cvtSdwaVOP2b";
-        }
+        foreach _ = BoolToList<P.HasExtSDWA>.ret in
+          def _sdwa  : VOP2_SDWA_Pseudo <opName, P> {
+            let AsmMatchConverter = "cvtSdwaVOP2b";
+          }
         foreach _ = BoolToList<P.HasExtDPP>.ret in
           def _dpp  : VOP2_DPP_Pseudo <opName, P>;
       }
@@ -220,9 +222,10 @@ multiclass VOP2eInst <string opName,
       def _e32 : VOP2_Pseudo <opName, P>,
                  Commutable_REV<revOp#"_e32", !eq(revOp, opName)>;
 
-      def _sdwa : VOP2_SDWA_Pseudo <opName, P> {
-        let AsmMatchConverter = "cvtSdwaVOP2b";
-      }
+      foreach _ = BoolToList<P.HasExtSDWA>.ret in
+        def _sdwa : VOP2_SDWA_Pseudo <opName, P> {
+          let AsmMatchConverter = "cvtSdwaVOP2b";
+        }
 
       foreach _ = BoolToList<P.HasExtDPP>.ret in
         def _dpp  : VOP2_DPP_Pseudo <opName, P>;
@@ -882,6 +885,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
       VOP3e_gfx10<{0, 1, 0, 0, op{5-0}}, !cast<VOP3_Pseudo>(NAME#"_e64").Pfl>;
   }
   multiclass VOP2_Real_sdwa_gfx10<bits<6> op> {
+    foreach _ = BoolToList<!cast<VOP2_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
     def _sdwa_gfx10 :
       VOP_SDWA10_Real<!cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa")>,
       VOP2_SDWA9Ae<op{5-0}, !cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa").Pfl> {
@@ -924,6 +928,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
   let DecoderNamespace = "SDWA10" in {
     multiclass VOP2_Real_sdwa_gfx10_with_name<bits<6> op, string opName,
                                               string asmName> {
+      foreach _ = BoolToList<!cast<VOP2_Pseudo>(opName#"_e32").Pfl.HasExtSDWA9>.ret in
       def _sdwa_gfx10 :
         VOP_SDWA10_Real<!cast<VOP2_SDWA_Pseudo>(opName#"_sdwa")>,
         VOP2_SDWA9Ae<op{5-0}, !cast<VOP2_SDWA_Pseudo>(opName#"_sdwa").Pfl> {
@@ -965,6 +970,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
         VOP3_Pseudo Ps = !cast<VOP3_Pseudo>(opName#"_e64");
         let AsmString = asmName # Ps.AsmOperands;
       }
+    foreach _ = BoolToList<!cast<VOP2_Pseudo>(opName#"_e32").Pfl.HasExtSDWA9>.ret in
     def _sdwa_gfx10 :
       VOP_SDWA10_Real<!cast<VOP2_SDWA_Pseudo>(opName#"_sdwa")>,
       VOP2_SDWA9Ae<op{5-0}, !cast<VOP2_SDWA_Pseudo>(opName#"_sdwa").Pfl> {
@@ -988,6 +994,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
       }
 
     let WaveSizePredicate = isWave32 in {
+      foreach _ = BoolToList<!cast<VOP2_Pseudo>(opName#"_e32").Pfl.HasExtSDWA9>.ret in
       def _sdwa_w32_gfx10 :
         Base_VOP_SDWA10_Real<!cast<VOP2_SDWA_Pseudo>(opName#"_sdwa")>,
         VOP2_SDWA9Ae<op{5-0}, !cast<VOP2_SDWA_Pseudo>(opName#"_sdwa").Pfl> {
@@ -1011,6 +1018,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
     } // End WaveSizePredicate = isWave32
 
     let WaveSizePredicate = isWave64 in {
+      foreach _ = BoolToList<!cast<VOP2_Pseudo>(opName#"_e32").Pfl.HasExtSDWA9>.ret in
       def _sdwa_w64_gfx10 :
         Base_VOP_SDWA10_Real<!cast<VOP2_SDWA_Pseudo>(opName#"_sdwa")>,
         VOP2_SDWA9Ae<op{5-0}, !cast<VOP2_SDWA_Pseudo>(opName#"_sdwa").Pfl> {
@@ -1286,12 +1294,14 @@ multiclass Base_VOP2_Real_e32e64_vi <bits<6> op> :
 } // End AssemblerPredicates = [isGFX8GFX9], DecoderNamespace = "GFX8"
 
 multiclass VOP2_SDWA_Real <bits<6> op> {
+  foreach _ = BoolToList<!cast<VOP2_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA>.ret in
   def _sdwa_vi :
     VOP_SDWA_Real <!cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOP2_SDWAe <op{5-0}, !cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
 }
 
 multiclass VOP2_SDWA9_Real <bits<6> op> {
+  foreach _ = BoolToList<!cast<VOP2_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
   def _sdwa_gfx9 :
     VOP_SDWA9_Real <!cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOP2_SDWA9Ae <op{5-0}, !cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
@@ -1314,12 +1324,13 @@ multiclass VOP2be_Real_e32e64_vi_only <bits<6> op, string OpName, string AsmName
       let AsmString = AsmName # ps.AsmOperands;
       let DecoderNamespace = "GFX8";
     }
-  def _sdwa_vi :
-    VOP_SDWA_Real <!cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa")>,
-    VOP2_SDWAe <op{5-0}, !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa").Pfl> {
-      VOP2_SDWA_Pseudo ps = !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa");
-      let AsmString = AsmName # ps.AsmOperands;
-    }
+  foreach _ = BoolToList<!cast<VOP2_Pseudo>(OpName#"_e32").Pfl.HasExtSDWA>.ret in
+    def _sdwa_vi :
+      VOP_SDWA_Real <!cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa")>,
+      VOP2_SDWAe <op{5-0}, !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa").Pfl> {
+        VOP2_SDWA_Pseudo ps = !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa");
+        let AsmString = AsmName # ps.AsmOperands;
+      }
   foreach _ = BoolToList<!cast<VOP2_Pseudo>(OpName#"_e32").Pfl.HasExtDPP>.ret in
     def _dpp_vi :
       VOP_DPP_Real<!cast<VOP2_DPP_Pseudo>(OpName#"_dpp"), SIEncodingFamily.VI>,
@@ -1347,12 +1358,13 @@ multiclass VOP2be_Real_e32e64_gfx9 <bits<6> op, string OpName, string AsmName> {
       let AsmString = AsmName # ps.AsmOperands;
       let DecoderNamespace = "GFX9";
     }
-  def _sdwa_gfx9 :
-    VOP_SDWA9_Real <!cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa")>,
-    VOP2_SDWA9Ae <op{5-0}, !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa").Pfl> {
-      VOP2_SDWA_Pseudo ps = !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa");
-      let AsmString = AsmName # ps.AsmOperands;
-    }
+  foreach _ = BoolToList<!cast<VOP2_Pseudo>(OpName#"_e32").Pfl.HasExtSDWA9>.ret in
+    def _sdwa_gfx9 :
+      VOP_SDWA9_Real <!cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa")>,
+      VOP2_SDWA9Ae <op{5-0}, !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa").Pfl> {
+        VOP2_SDWA_Pseudo ps = !cast<VOP2_SDWA_Pseudo>(OpName#"_sdwa");
+        let AsmString = AsmName # ps.AsmOperands;
+      }
   foreach _ = BoolToList<!cast<VOP2_Pseudo>(OpName#"_e32").Pfl.HasExtDPP>.ret in
     def _dpp_gfx9 :
       VOP_DPP_Real<!cast<VOP2_DPP_Pseudo>(OpName#"_dpp"), SIEncodingFamily.GFX9>,
@@ -1374,10 +1386,11 @@ multiclass VOP2_Real_e32e64_gfx9 <bits<6> op> {
     VOP3e_vi <{0, 1, 0, 0, op{5-0}}, !cast<VOP3_Pseudo>(NAME#"_e64").Pfl> {
       let DecoderNamespace = "GFX9";
     }
-  def _sdwa_gfx9 :
-    VOP_SDWA9_Real <!cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa")>,
-    VOP2_SDWA9Ae <op{5-0}, !cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa").Pfl> {
-    }
+  foreach _ = BoolToList<!cast<VOP2_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
+    def _sdwa_gfx9 :
+      VOP_SDWA9_Real <!cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa")>,
+      VOP2_SDWA9Ae <op{5-0}, !cast<VOP2_SDWA_Pseudo>(NAME#"_sdwa").Pfl> {
+      }
   foreach _ = BoolToList<!cast<VOP2_Pseudo>(NAME#"_e32").Pfl.HasExtDPP>.ret in
     def _dpp_gfx9 :
       VOP_DPP_Real<!cast<VOP2_DPP_Pseudo>(NAME#"_dpp"), SIEncodingFamily.GFX9>,

diff --git a/lib/Target/AMDGPU/VOPCInstructions.td b/lib/Target/AMDGPU/VOPCInstructions.td
index e011b0a447edd3a20a8c7b540837792d6924a5bf..8ef0ec7b71f4989415c83c6813c26f8f90644710 100644 (file)
--- a/lib/Target/AMDGPU/VOPCInstructions.td
+++ b/lib/Target/AMDGPU/VOPCInstructions.td
@@ -225,6 +225,7 @@ multiclass VOPC_Pseudos <string opName,
     let isCommutable = 1;
   }
 
+  foreach _ = BoolToList<P.HasExtSDWA>.ret in
   def _sdwa : VOPC_SDWA_Pseudo <opName, P> {
     let Defs = !if(DefExec, [VCC, EXEC], [VCC]);
     let SchedRW = P.Schedule;
@@ -261,6 +262,7 @@ multiclass VOPCX_Pseudos <string opName,
     let SubtargetPredicate = HasNoSdstCMPX;
   }
 
+  foreach _ = BoolToList<P_NoSDst.HasExtSDWA>.ret in
   def _nosdst_sdwa : VOPC_SDWA_Pseudo <opName#"_nosdst", P_NoSDst> {
     let Defs = [EXEC];
     let SchedRW = P_NoSDst.Schedule;
@@ -670,6 +672,7 @@ multiclass VOPC_Class_Pseudos <string opName, VOPC_Profile p, bit DefExec,
     let SchedRW = p.Schedule;
   }
 
+  foreach _ = BoolToList<p.HasExtSDWA>.ret in
   def _sdwa : VOPC_SDWA_Pseudo <opName, p> {
     let Defs = !if(DefExec, !if(DefVcc, [VCC, EXEC], [EXEC]),
                             !if(DefVcc, [VCC], []));
@@ -699,6 +702,7 @@ multiclass VOPCX_Class_Pseudos <string opName,
     let SubtargetPredicate = HasNoSdstCMPX;
   }
 
+  foreach _ = BoolToList<P_NoSDst.HasExtSDWA>.ret in
   def _nosdst_sdwa : VOPC_SDWA_Pseudo <opName#"_nosdst", P_NoSDst> {
     let Defs = [EXEC];
     let SchedRW = P_NoSDst.Schedule;
@@ -882,6 +886,7 @@ let AssemblerPredicate = isGFX10Plus in {
       }
     } // End DecoderNamespace = "GFX10"
 
+    foreach _ = BoolToList<!cast<VOPC_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
     def _sdwa_gfx10 :
       VOP_SDWA10_Real<!cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa")>,
       VOPC_SDWA9e<op{7-0}, !cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
@@ -907,6 +912,7 @@ let AssemblerPredicate = isGFX10Plus in {
         }
     } // End DecoderNamespace = "GFX10"
 
+    foreach _ = BoolToList<!cast<VOPC_Pseudo>(NAME#"_nosdst_e32").Pfl.HasExtSDWA9>.ret in
     def _sdwa_gfx10 :
       VOP_SDWA10_Real<!cast<VOPC_SDWA_Pseudo>(NAME#"_nosdst_sdwa")>,
       VOPC_SDWA9e<op{7-0}, !cast<VOPC_SDWA_Pseudo>(NAME#"_nosdst_sdwa").Pfl> {
@@ -1227,10 +1233,12 @@ multiclass VOPC_Real_vi <bits<10> op> {
     }
   }
 
+  foreach _ = BoolToList<!cast<VOPC_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA>.ret in
   def _sdwa_vi :
     VOP_SDWA_Real <!cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOPC_SDWAe <op{7-0}, !cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
 
+  foreach _ = BoolToList<!cast<VOPC_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
   def _sdwa_gfx9 :
     VOP_SDWA9_Real <!cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa")>,
     VOPC_SDWA9e <op{7-0}, !cast<VOPC_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;