[X86][AVX] Cleanup AVX2 vector integer truncation costs

Noticed while investigating PR50364, the truncation costs for v4i64->v4i16/v4i8 and v8i32->v8i8 were way too optimistic for a shuffle sequence that usually matches the AVX1 codegen (they matched AVX512 numbers which have actual truncation instructions!).

Author: RKSimon

llvm/lib/Target/X86/X86TargetTransformInfo.cpp

@@ -1684,6 +1684,9 @@ InstructionCost X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
     { ISD::TRUNCATE,  MVT::v2i1,    MVT::v2i64,  2 }, // vpsllq+vptestmq
     { ISD::TRUNCATE,  MVT::v4i1,    MVT::v4i64,  2 }, // vpsllq+vptestmq
     { ISD::TRUNCATE,  MVT::v4i32,   MVT::v4i64,  1 }, // vpmovqd
+    { ISD::TRUNCATE,  MVT::v4i8,    MVT::v4i64,  2 }, // vpmovqb
+    { ISD::TRUNCATE,  MVT::v4i16,   MVT::v4i64,  2 }, // vpmovqw
+    { ISD::TRUNCATE,  MVT::v8i8,    MVT::v8i32,  2 }, // vpmovwb
 
     // sign extend is vpcmpeq+maskedmove+vpmovdw+vpacksswb
     // zero extend is vpcmpeq+maskedmove+vpmovdw+vpsrlw+vpackuswb
@@ -1773,10 +1776,6 @@ InstructionCost X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
 
     { ISD::TRUNCATE,    MVT::v4i32,  MVT::v4i64,  2 },
     { ISD::TRUNCATE,    MVT::v8i1,   MVT::v8i32,  2 },
-
-    { ISD::TRUNCATE,    MVT::v4i8,   MVT::v4i64,  2 },
-    { ISD::TRUNCATE,    MVT::v4i16,  MVT::v4i64,  2 },
-    { ISD::TRUNCATE,    MVT::v8i8,   MVT::v8i32,  2 },
     { ISD::TRUNCATE,    MVT::v8i16,  MVT::v8i32,  2 },
 
     { ISD::FP_EXTEND,   MVT::v8f64,  MVT::v8f32,  3 },

llvm/test/Analysis/CostModel/X86/cast.ll

@@ -183,8 +183,8 @@ define i32 @zext_sext(<8 x i1> %in) {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %E = trunc <4 x i64> undef to <4 x i32>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F = trunc <8 x i32> undef to <8 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F1 = trunc <16 x i16> undef to <16 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F2 = trunc <8 x i32> undef to <8 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F3 = trunc <4 x i64> undef to <4 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F2 = trunc <8 x i32> undef to <8 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F3 = trunc <4 x i64> undef to <4 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %G = trunc <8 x i64> undef to <8 x i32>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %G1 = trunc <16 x i32> undef to <16 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %G2 = trunc <16 x i32> undef to <16 x i8>

llvm/test/Analysis/CostModel/X86/trunc.ll

@@ -108,7 +108,7 @@ define i32 @trunc_vXi16() {
 ; AVX2-LABEL: 'trunc_vXi16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i16>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
@@ -239,53 +239,29 @@ define i32 @trunc_vXi8() {
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
-; AVX1-LABEL: 'trunc_vXi8'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i8
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i32 = trunc i32 undef to i8
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i16 = trunc i16 undef to i8
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
-;
-; AVX2-LABEL: 'trunc_vXi8'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i8
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i32 = trunc i32 undef to i8
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i16 = trunc i16 undef to i8
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+; AVX-LABEL: 'trunc_vXi8'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i32 = trunc i32 undef to i8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i16 = trunc i16 undef to i8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'trunc_vXi8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i8
@@ -334,30 +310,6 @@ define i32 @trunc_vXi8() {
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
-;
-; BTVER2-LABEL: 'trunc_vXi8'
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i64 = trunc i64 undef to i8
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i32 = trunc i32 undef to i8
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %i16 = trunc i16 undef to i8
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64i16 = trunc <64 x i16> undef to <64 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %i64 = trunc i64 undef to i8
   %V2i64 = trunc <2 x i64> undef to <2 x i8>