52 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py2; RUN: opt -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=thumbv8m.main < %s | FileCheck %s --check-prefix=V8M3; RUN: opt -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=armv8a-linux-gnueabihf < %s | FileCheck %s --check-prefix=NEON4; RUN: opt -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=armv8.1m.main -mattr=+mve < %s | FileCheck %s --check-prefix=MVE5 6target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"7 8define i32 @reduce_i1(i32 %arg) {9; V8M-LABEL: 'reduce_i1'10; V8M-NEXT: Cost Model: Found costs of 1 for: %V1 = call i1 @llvm.vector.reduce.or.v1i1(<1 x i1> undef)11; V8M-NEXT: Cost Model: Found costs of RThru:3 CodeSize:2 Lat:2 SizeLat:2 for: %V2 = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> undef)12; V8M-NEXT: Cost Model: Found costs of RThru:5 CodeSize:2 Lat:2 SizeLat:2 for: %V4 = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> undef)13; V8M-NEXT: Cost Model: Found costs of RThru:9 CodeSize:2 Lat:2 SizeLat:2 for: %V8 = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> undef)14; V8M-NEXT: Cost Model: Found costs of RThru:17 CodeSize:2 Lat:2 SizeLat:2 for: %V16 = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> undef)15; V8M-NEXT: Cost Model: Found costs of RThru:33 CodeSize:2 Lat:2 SizeLat:2 for: %V32 = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> undef)16; V8M-NEXT: Cost Model: Found costs of RThru:65 CodeSize:2 Lat:2 SizeLat:2 for: %V64 = call i1 @llvm.vector.reduce.or.v64i1(<64 x i1> undef)17; V8M-NEXT: Cost Model: Found costs of RThru:129 CodeSize:2 Lat:2 SizeLat:2 for: %V128 = call i1 @llvm.vector.reduce.or.v128i1(<128 x i1> undef)18; V8M-NEXT: Cost Model: Found costs of 1 for: ret i32 undef19;20; NEON-LABEL: 'reduce_i1'21; NEON-NEXT: Cost Model: Found costs of 3 for: %V1 = call i1 @llvm.vector.reduce.or.v1i1(<1 x i1> undef)22; NEON-NEXT: Cost Model: Found costs of 2 for: %V2 = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> undef)23; NEON-NEXT: Cost Model: Found costs of 2 for: %V4 = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> undef)24; NEON-NEXT: Cost Model: Found costs of 2 for: %V8 = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> undef)25; NEON-NEXT: Cost Model: Found costs of 2 for: %V16 = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> undef)26; NEON-NEXT: Cost Model: Found costs of RThru:97 CodeSize:2 Lat:2 SizeLat:2 for: %V32 = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> undef)27; NEON-NEXT: Cost Model: Found costs of RThru:193 CodeSize:2 Lat:2 SizeLat:2 for: %V64 = call i1 @llvm.vector.reduce.or.v64i1(<64 x i1> undef)28; NEON-NEXT: Cost Model: Found costs of RThru:385 CodeSize:2 Lat:2 SizeLat:2 for: %V128 = call i1 @llvm.vector.reduce.or.v128i1(<128 x i1> undef)29; NEON-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef30;31; MVE-LABEL: 'reduce_i1'32; MVE-NEXT: Cost Model: Found costs of 4 for: %V1 = call i1 @llvm.vector.reduce.or.v1i1(<1 x i1> undef)33; MVE-NEXT: Cost Model: Found costs of RThru:3 CodeSize:2 Lat:2 SizeLat:2 for: %V2 = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> undef)34; MVE-NEXT: Cost Model: Found costs of RThru:3 CodeSize:2 Lat:2 SizeLat:2 for: %V4 = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> undef)35; MVE-NEXT: Cost Model: Found costs of RThru:3 CodeSize:2 Lat:2 SizeLat:2 for: %V8 = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> undef)36; MVE-NEXT: Cost Model: Found costs of RThru:3 CodeSize:2 Lat:2 SizeLat:2 for: %V16 = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> undef)37; MVE-NEXT: Cost Model: Found costs of RThru:257 CodeSize:2 Lat:2 SizeLat:2 for: %V32 = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> undef)38; MVE-NEXT: Cost Model: Found costs of RThru:513 CodeSize:2 Lat:2 SizeLat:2 for: %V64 = call i1 @llvm.vector.reduce.or.v64i1(<64 x i1> undef)39; MVE-NEXT: Cost Model: Found costs of RThru:1025 CodeSize:2 Lat:2 SizeLat:2 for: %V128 = call i1 @llvm.vector.reduce.or.v128i1(<128 x i1> undef)40; MVE-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef41;42 %V1 = call i1 @llvm.vector.reduce.or.v1i1(<1 x i1> undef)43 %V2 = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> undef)44 %V4 = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> undef)45 %V8 = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> undef)46 %V16 = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> undef)47 %V32 = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> undef)48 %V64 = call i1 @llvm.vector.reduce.or.v64i1(<64 x i1> undef)49 %V128 = call i1 @llvm.vector.reduce.or.v128i1(<128 x i1> undef)50 ret i32 undef51}52