|
| 1 | +; This file is a part of Julia. License is MIT: https://julialang.org/license |
| 2 | + |
| 3 | +; RUN: opt -enable-new-pm=1 --opaque-pointers=0 --load-pass-plugin=libjulia-codegen%shlibext -passes='JuliaMultiVersioning,CPUFeatures' -S %s | FileCheck %s |
| 4 | + |
| 5 | +; RUN: opt -enable-new-pm=1 --opaque-pointers=1 --load-pass-plugin=libjulia-codegen%shlibext -passes='JuliaMultiVersioning,CPUFeatures' -S %s | FileCheck %s |
| 6 | + |
| 7 | +; COM: This test checks that multiversioning actually happens from start to finish |
| 8 | +; COM: We need the fvars for a proper test |
| 9 | + |
| 10 | + |
| 11 | + |
| 12 | +; CHECK: @jl_fvar_idxs = hidden constant [5 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4], align 16 |
| 13 | +; CHECK: @jl_gvar_idxs = hidden constant [0 x i32] zeroinitializer, align 16 |
| 14 | +; TYPED: @simd_test.reloc_slot = hidden global i32 (<4 x i32>)* null |
| 15 | +; OPAQUE: @simd_test.reloc_slot = hidden global ptr null |
| 16 | +; TYPED: @jl_fvar_offsets = hidden constant [6 x i32] [i32 5, i32 0, i32 trunc (i64 sub (i64 ptrtoint (float (float, float)* @fastmath_test to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (i32)* @loop_test to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test_call to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32)] |
| 17 | +; OPAQUE: @jl_fvar_offsets = hidden constant [6 x i32] [i32 5, i32 0, i32 trunc (i64 sub (i64 ptrtoint (ptr @fastmath_test to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @loop_test to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test_call to i64), i64 ptrtoint (ptr @boring to i64)) to i32)] |
| 18 | +; CHECK: @jl_gvar_base = hidden constant i64 0 |
| 19 | +; CHECK: @jl_gvar_offsets = hidden constant [1 x i32] zeroinitializer |
| 20 | +; TYPED: @jl_clone_slots = hidden constant [3 x i32] [i32 1, i32 3, i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)** @simd_test.reloc_slot to i64), i64 ptrtoint (i64* @jl_gvar_base to i64)) to i32)] |
| 21 | +; OPAQUE: @jl_clone_slots = hidden constant [3 x i32] [i32 1, i32 3, i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test.reloc_slot to i64), i64 ptrtoint (ptr @jl_gvar_base to i64)) to i32)] |
| 22 | +; CHECK: @jl_clone_idxs = hidden constant [10 x i32] [i32 -2147483647, i32 3, i32 -2147483647, i32 3, i32 4, i32 1, i32 1, i32 2, i32 -2147483645, i32 4] |
| 23 | +; TYPED: @jl_clone_offsets = hidden constant [9 x i32] [i32 trunc (i64 sub (i64 ptrtoint (i32 (i32)* @boring.1 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (float (float, float)* @fastmath_test.1 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (i32)* @loop_test.1 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test.1 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test_call.1 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (float (float, float)* @fastmath_test.2 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (i32)* @loop_test.2 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test.2 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (i32 (<4 x i32>)* @simd_test_call.2 to i64), i64 ptrtoint (i32 (i32)* @boring to i64)) to i32)] |
| 24 | +; OPAQUE: @jl_clone_offsets = hidden constant [9 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr @boring.1 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @fastmath_test.1 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @loop_test.1 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test.1 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test_call.1 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @fastmath_test.2 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @loop_test.2 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test.2 to i64), i64 ptrtoint (ptr @boring to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @simd_test_call.2 to i64), i64 ptrtoint (ptr @boring to i64)) to i32)] |
| 25 | +; TYPED: @jl_fvar_base = hidden alias i64, bitcast (i32 (i32)* @boring to i64*) |
| 26 | +; OPAQUE: @jl_fvar_base = hidden alias i64, ptr @boring |
| 27 | + |
| 28 | + |
| 29 | +target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128-ni:10:11:12:13" |
| 30 | +target triple = "x86_64-linux-gnu" |
| 31 | + |
| 32 | +@jl_fvars = global [5 x i64*] [i64* bitcast (i32 (i32)* @boring to i64*), |
| 33 | + i64* bitcast (float (float, float)* @fastmath_test to i64*), |
| 34 | + i64* bitcast (i32 (i32)* @loop_test to i64*), |
| 35 | + i64* bitcast (i32 (<4 x i32>)* @simd_test to i64*), |
| 36 | + i64* bitcast (i32 (<4 x i32>)* @simd_test_call to i64*) |
| 37 | + ], align 16 |
| 38 | +@jl_gvars = global [0 x i64*] zeroinitializer, align 16 |
| 39 | +@jl_fvar_idxs = hidden constant [5 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4], align 16 |
| 40 | +@jl_gvar_idxs = hidden constant [0 x i32] zeroinitializer, align 16 |
| 41 | + |
| 42 | +declare i1 @julia.cpu.have_fma.f32() |
| 43 | + |
| 44 | +; CHECK: @boring{{.*}}#[[BORING_BASE:[0-9]+]] |
| 45 | +define noundef i32 @boring(i32 noundef %0) { |
| 46 | + ret i32 %0 |
| 47 | +} |
| 48 | + |
| 49 | +; CHECK: @fastmath_test{{.*}}#[[NOT_BORING_BASE:[0-9]+]] |
| 50 | +; CHECK: %3 = sitofp i1 false to float |
| 51 | +define noundef float @fastmath_test(float noundef %0, float noundef %1) { |
| 52 | + %3 = call i1 @julia.cpu.have_fma.f32() |
| 53 | + %4 = sitofp i1 %3 to float |
| 54 | + %5 = fadd fast float %0, %4 |
| 55 | + ret float %5 |
| 56 | +} |
| 57 | + |
| 58 | +; CHECK: @loop_test{{.*}}#[[NOT_BORING_BASE:[0-9]+]] |
| 59 | +define noundef i32 @loop_test(i32 noundef %0) { |
| 60 | + %2 = icmp sgt i32 %0, 0 |
| 61 | + br i1 %2, label %5, label %3 |
| 62 | + |
| 63 | +3: ; preds = %5, %1 |
| 64 | + %4 = phi i32 [ 0, %1 ], [ %9, %5 ] |
| 65 | + ret i32 %4 |
| 66 | + |
| 67 | +5: ; preds = %1, %5 |
| 68 | + %6 = phi i32 [ %10, %5 ], [ 0, %1 ] |
| 69 | + %7 = phi i32 [ %9, %5 ], [ 0, %1 ] |
| 70 | + %8 = lshr i32 %6, 1 |
| 71 | + %9 = add nuw nsw i32 %8, %7 |
| 72 | + %10 = add nuw nsw i32 %6, 1 |
| 73 | + %11 = icmp eq i32 %10, %0 |
| 74 | + br i1 %11, label %3, label %5;, !llvm.loop - |
| 75 | +} |
| 76 | + |
| 77 | +; CHECK: @simd_test{{.*}}#[[SIMD_BASE_RELOC:[0-9]+]] |
| 78 | +define noundef i32 @simd_test(<4 x i32> noundef %0) { |
| 79 | + %2 = extractelement <4 x i32> %0, i64 0 |
| 80 | + ret i32 %2 |
| 81 | +} |
| 82 | + |
| 83 | +; CHECK: @simd_test_call{{.*}}#[[NOT_BORING_BASE:[0-9]+]] |
| 84 | +define noundef i32 @simd_test_call(<4 x i32> noundef %0) { |
| 85 | + %2 = call noundef i32 @simd_test(<4 x i32> noundef %0) |
| 86 | + ret i32 %2 |
| 87 | +} |
| 88 | + |
| 89 | +; CHECK: @boring{{.*}}#[[BORING_CLONE:[0-9]+]] |
| 90 | + |
| 91 | +; CHECK: @fastmath_test{{.*}}#[[NOT_BORING_CLONE1:[0-9]+]] |
| 92 | +; CHECK: %3 = sitofp i1 false to float |
| 93 | + |
| 94 | +; CHECK: @fastmath_test{{.*}}#[[NOT_BORING_CLONE2:[0-9]+]] |
| 95 | +; CHECK: %3 = sitofp i1 true to float |
| 96 | + |
| 97 | +; CHECK: @loop_test{{.*}}#[[NOT_BORING_CLONE1:[0-9]+]] |
| 98 | + |
| 99 | +; CHECK: @loop_test{{.*}}#[[NOT_BORING_CLONE2:[0-9]+]] |
| 100 | + |
| 101 | +; CHECK: @simd_test{{.*}}#[[SIMD_CLONE1:[0-9]+]] |
| 102 | + |
| 103 | +; CHECK: @simd_test{{.*}}#[[SIMD_CLONE2:[0-9]+]] |
| 104 | + |
| 105 | +; CHECK: @simd_test_call{{.*}}#[[NOT_BORING_CLONE1:[0-9]+]] |
| 106 | +; TYPED: %2 = load i32 (<4 x i32>)*, i32 (<4 x i32>)** @simd_test.reloc_slot, align 8, !tbaa !8, !invariant.load !12 |
| 107 | +; OPAQUE: %2 = load ptr, ptr @simd_test.reloc_slot, align 8, !tbaa !8, !invariant.load !12 |
| 108 | +; CHECK: %3 = call noundef i32 %2(<4 x i32> noundef %0) |
| 109 | + |
| 110 | +; CHECK: @simd_test_call{{.*}}#[[NOT_BORING_CLONE2:[0-9]+]] |
| 111 | +; CHECK: %2 = call noundef i32 @simd_test.2(<4 x i32> noundef %0) |
| 112 | + |
| 113 | +; CHECK-DAG: attributes #[[BORING_BASE]] = { "julia.mv.clone"="0" "julia.mv.clones"="2" "julia.mv.fvar" "target-cpu"="x86-64" "target-features"="+cx16,-sse3,-pclmul,-ssse3,-fma,-sse4.1,-sse4.2,-movbe,-popcnt,-aes,-xsave,-avx,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sahf,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 114 | +; CHECK-DAG: attributes #[[NOT_BORING_BASE]] = { "julia.mv.clone"="0" "julia.mv.clones"="6" "julia.mv.fvar" "target-cpu"="x86-64" "target-features"="+cx16,-sse3,-pclmul,-ssse3,-fma,-sse4.1,-sse4.2,-movbe,-popcnt,-aes,-xsave,-avx,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sahf,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 115 | +; CHECK-DAG: attributes #[[SIMD_BASE_RELOC]] = { "julia.mv.clone"="0" "julia.mv.clones"="6" "julia.mv.reloc" "target-cpu"="x86-64" "target-features"="+cx16,-sse3,-pclmul,-ssse3,-fma,-sse4.1,-sse4.2,-movbe,-popcnt,-aes,-xsave,-avx,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sahf,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 116 | +; CHECK-DAG: attributes #[[BORING_CLONE]] = { "julia.mv.clone"="1" "julia.mv.clones"="2" "julia.mv.fvar" "target-cpu"="sandybridge" "target-features"="+sahf,+avx,+xsave,+popcnt,+sse4.2,+sse4.1,+cx16,+ssse3,+pclmul,+sse3,-fma,-movbe,-aes,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 117 | +; CHECK-DAG: attributes #[[NOT_BORING_CLONE1]] = { "julia.mv.clone"="1" "julia.mv.clones"="6" "julia.mv.fvar" "target-cpu"="sandybridge" "target-features"="+sahf,+avx,+xsave,+popcnt,+sse4.2,+sse4.1,+cx16,+ssse3,+pclmul,+sse3,-fma,-movbe,-aes,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 118 | +; CHECK-DAG: attributes #[[NOT_BORING_CLONE2]] = { "julia.mv.clone"="2" "julia.mv.clones"="6" "julia.mv.fvar" "target-cpu"="haswell" "target-features"="+lzcnt,+sahf,+bmi2,+avx2,+bmi,+fsgsbase,+f16c,+avx,+xsave,+popcnt,+movbe,+sse4.2,+sse4.1,+cx16,+fma,+ssse3,+pclmul,+sse3,-aes,-rdrnd,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 119 | +; CHECK-DAG: attributes #[[SIMD_CLONE1]] = { "julia.mv.clone"="1" "julia.mv.clones"="6" "julia.mv.reloc" "target-cpu"="sandybridge" "target-features"="+sahf,+avx,+xsave,+popcnt,+sse4.2,+sse4.1,+cx16,+ssse3,+pclmul,+sse3,-fma,-movbe,-aes,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 120 | +; CHECK-DAG: attributes #[[SIMD_CLONE2]] = { "julia.mv.clone"="2" "julia.mv.clones"="6" "julia.mv.reloc" "target-cpu"="haswell" "target-features"="+lzcnt,+sahf,+bmi2,+avx2,+bmi,+fsgsbase,+f16c,+avx,+xsave,+popcnt,+movbe,+sse4.2,+sse4.1,+cx16,+fma,+ssse3,+pclmul,+sse3,-aes,-rdrnd,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8" } |
| 121 | + |
| 122 | + |
| 123 | +!llvm.module.flags = !{!0, !2} |
| 124 | + |
| 125 | + |
| 126 | +!0 = !{i32 1, !"julia.mv.enable", i32 1} |
| 127 | +!1 = !{!1} |
| 128 | +!2 = !{i32 1, !"julia.mv.specs", !3} |
| 129 | +!3 = !{!4, !5, !6} |
| 130 | +!4 = !{!"x86-64", !"+cx16,-sse3,-pclmul,-ssse3,-fma,-sse4.1,-sse4.2,-movbe,-popcnt,-aes,-xsave,-avx,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sahf,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8", i32 0, i32 0} |
| 131 | +!5 = !{!"sandybridge", !"+sahf,+avx,+xsave,+popcnt,+sse4.2,+sse4.1,+cx16,+ssse3,+pclmul,+sse3,-fma,-movbe,-aes,-f16c,-rdrnd,-fsgsbase,-bmi,-avx2,-bmi2,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-lzcnt,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8", i32 0, i32 2} |
| 132 | +!6 = !{!"haswell", !"+lzcnt,+sahf,+bmi2,+avx2,+bmi,+fsgsbase,+f16c,+avx,+xsave,+popcnt,+movbe,+sse4.2,+sse4.1,+cx16,+fma,+ssse3,+pclmul,+sse3,-aes,-rdrnd,-rtm,-avx512f,-avx512dq,-rdseed,-adx,-avx512ifma,-clflushopt,-clwb,-avx512pf,-avx512er,-avx512cd,-sha,-avx512bw,-avx512vl,-prefetchwt1,-avx512vbmi,-pku,-waitpkg,-avx512vbmi2,-shstk,-gfni,-vaes,-vpclmulqdq,-avx512vnni,-avx512bitalg,-avx512vpopcntdq,-rdpid,-cldemote,-movdiri,-movdir64b,-enqcmd,-uintr,-avx512vp2intersect,-serialize,-tsxldtrk,-pconfig,-amx-bf16,-avx512fp16,-amx-tile,-amx-int8,-sse4a,-prfchw,-xop,-fma4,-tbm,-mwaitx,-xsaveopt,-xsavec,-xsaves,-clzero,-wbnoinvd,-avxvnni,-avx512bf16,-ptwrite,+sse2,+mmx,+fxsr,+64bit,+cx8", i32 1, i32 284} |
0 commit comments