From dee4d699b27de6d828c1e1fa3e6f5cb655f89e99 Mon Sep 17 00:00:00 2001 From: Roman Lebedev Date: Mon, 4 Oct 2021 20:05:50 +0300 Subject: [PATCH] [NFC][X86][LV] Add costmodel test coverage for interleaved i64/f64 load/store stride=6 --- .../X86/interleaved-load-f64-stride-6.ll | 83 ++++++++++++++++++ .../X86/interleaved-load-i64-stride-6.ll | 83 ++++++++++++++++++ .../X86/interleaved-store-f64-stride-6.ll | 84 +++++++++++++++++++ .../X86/interleaved-store-i64-stride-6.ll | 84 +++++++++++++++++++ 4 files changed, 334 insertions(+) create mode 100644 llvm/test/Analysis/CostModel/X86/interleaved-load-f64-stride-6.ll create mode 100644 llvm/test/Analysis/CostModel/X86/interleaved-load-i64-stride-6.ll create mode 100644 llvm/test/Analysis/CostModel/X86/interleaved-store-f64-stride-6.ll create mode 100644 llvm/test/Analysis/CostModel/X86/interleaved-store-i64-stride-6.ll diff --git a/llvm/test/Analysis/CostModel/X86/interleaved-load-f64-stride-6.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-f64-stride-6.ll new file mode 100644 index 000000000000..9ff11c8427f6 --- /dev/null +++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-f64-stride-6.ll @@ -0,0 +1,83 @@ +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,SSE2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX1 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw,+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX512 +; REQUIRES: asserts + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@A = global [1024 x double] zeroinitializer, align 128 +@B = global [1024 x i8] zeroinitializer, align 128 + +; CHECK: LV: Checking a loop in "test" +; +; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load double, double* %in0, align 8 +; SSE2: LV: Found an estimated cost of 18 for VF 2 For instruction: %v0 = load double, double* %in0, align 8 +; SSE2: LV: Found an estimated cost of 36 for VF 4 For instruction: %v0 = load double, double* %in0, align 8 +; +; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load double, double* %in0, align 8 +; AVX1: LV: Found an estimated cost of 21 for VF 2 For instruction: %v0 = load double, double* %in0, align 8 +; AVX1: LV: Found an estimated cost of 48 for VF 4 For instruction: %v0 = load double, double* %in0, align 8 +; AVX1: LV: Found an estimated cost of 96 for VF 8 For instruction: %v0 = load double, double* %in0, align 8 +; +; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load double, double* %in0, align 8 +; AVX2: LV: Found an estimated cost of 21 for VF 2 For instruction: %v0 = load double, double* %in0, align 8 +; AVX2: LV: Found an estimated cost of 48 for VF 4 For instruction: %v0 = load double, double* %in0, align 8 +; AVX2: LV: Found an estimated cost of 96 for VF 8 For instruction: %v0 = load double, double* %in0, align 8 +; +; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load double, double* %in0, align 8 +; AVX512: LV: Found an estimated cost of 11 for VF 2 For instruction: %v0 = load double, double* %in0, align 8 +; AVX512: LV: Found an estimated cost of 21 for VF 4 For instruction: %v0 = load double, double* %in0, align 8 +; AVX512: LV: Found an estimated cost of 51 for VF 8 For instruction: %v0 = load double, double* %in0, align 8 +; AVX512: LV: Found an estimated cost of 120 for VF 16 For instruction: %v0 = load double, double* %in0, align 8 +; AVX512: LV: Found an estimated cost of 240 for VF 32 For instruction: %v0 = load double, double* %in0, align 8 +; +; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load double, double* %in0, align 8 + +define void @test() { +entry: + br label %for.body + +for.body: + %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ] + + %iv.0 = add nuw nsw i64 %iv, 0 + %iv.1 = add nuw nsw i64 %iv, 1 + %iv.2 = add nuw nsw i64 %iv, 2 + %iv.3 = add nuw nsw i64 %iv, 3 + %iv.4 = add nuw nsw i64 %iv, 4 + %iv.5 = add nuw nsw i64 %iv, 5 + + %in0 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.0 + %in1 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.1 + %in2 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.2 + %in3 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.3 + %in4 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.4 + %in5 = getelementptr inbounds [1024 x double], [1024 x double]* @A, i64 0, i64 %iv.5 + + %v0 = load double, double* %in0 + %v1 = load double, double* %in1 + %v2 = load double, double* %in2 + %v3 = load double, double* %in3 + %v4 = load double, double* %in4 + %v5 = load double, double* %in5 + + %reduce.add.0 = fadd double %v0, %v1 + %reduce.add.1 = fadd double %reduce.add.0, %v2 + %reduce.add.2 = fadd double %reduce.add.1, %v3 + %reduce.add.3 = fadd double %reduce.add.2, %v4 + %reduce.add.4 = fadd double %reduce.add.3, %v5 + + %reduce.add.4.narrow = fptoui double %reduce.add.4 to i8 + + %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0 + store i8 %reduce.add.4.narrow, i8* %out + + %iv.next = add nuw nsw i64 %iv.0, 6 + %cmp = icmp ult i64 %iv.next, 1024 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.cond.cleanup: + ret void +} diff --git a/llvm/test/Analysis/CostModel/X86/interleaved-load-i64-stride-6.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i64-stride-6.ll new file mode 100644 index 000000000000..2e586d7d11f4 --- /dev/null +++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i64-stride-6.ll @@ -0,0 +1,83 @@ +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,SSE2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX1 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw,+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX512 +; REQUIRES: asserts + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@A = global [1024 x i64] zeroinitializer, align 128 +@B = global [1024 x i8] zeroinitializer, align 128 + +; CHECK: LV: Checking a loop in "test" +; +; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, i64* %in0, align 8 +; SSE2: LV: Found an estimated cost of 42 for VF 2 For instruction: %v0 = load i64, i64* %in0, align 8 +; SSE2: LV: Found an estimated cost of 84 for VF 4 For instruction: %v0 = load i64, i64* %in0, align 8 +; +; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX1: LV: Found an estimated cost of 33 for VF 2 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX1: LV: Found an estimated cost of 78 for VF 4 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX1: LV: Found an estimated cost of 156 for VF 8 For instruction: %v0 = load i64, i64* %in0, align 8 +; +; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX2: LV: Found an estimated cost of 33 for VF 2 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX2: LV: Found an estimated cost of 78 for VF 4 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX2: LV: Found an estimated cost of 156 for VF 8 For instruction: %v0 = load i64, i64* %in0, align 8 +; +; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX512: LV: Found an estimated cost of 11 for VF 2 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX512: LV: Found an estimated cost of 21 for VF 4 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX512: LV: Found an estimated cost of 51 for VF 8 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX512: LV: Found an estimated cost of 120 for VF 16 For instruction: %v0 = load i64, i64* %in0, align 8 +; AVX512: LV: Found an estimated cost of 240 for VF 32 For instruction: %v0 = load i64, i64* %in0, align 8 +; +; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i64, i64* %in0, align 8 + +define void @test() { +entry: + br label %for.body + +for.body: + %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ] + + %iv.0 = add nuw nsw i64 %iv, 0 + %iv.1 = add nuw nsw i64 %iv, 1 + %iv.2 = add nuw nsw i64 %iv, 2 + %iv.3 = add nuw nsw i64 %iv, 3 + %iv.4 = add nuw nsw i64 %iv, 4 + %iv.5 = add nuw nsw i64 %iv, 5 + + %in0 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.0 + %in1 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.1 + %in2 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.2 + %in3 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.3 + %in4 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.4 + %in5 = getelementptr inbounds [1024 x i64], [1024 x i64]* @A, i64 0, i64 %iv.5 + + %v0 = load i64, i64* %in0 + %v1 = load i64, i64* %in1 + %v2 = load i64, i64* %in2 + %v3 = load i64, i64* %in3 + %v4 = load i64, i64* %in4 + %v5 = load i64, i64* %in5 + + %reduce.add.0 = add i64 %v0, %v1 + %reduce.add.1 = add i64 %reduce.add.0, %v2 + %reduce.add.2 = add i64 %reduce.add.1, %v3 + %reduce.add.3 = add i64 %reduce.add.2, %v4 + %reduce.add.4 = add i64 %reduce.add.3, %v5 + + %reduce.add.4.narrow = trunc i64 %reduce.add.4 to i8 + + %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0 + store i8 %reduce.add.4.narrow, i8* %out + + %iv.next = add nuw nsw i64 %iv.0, 6 + %cmp = icmp ult i64 %iv.next, 1024 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.cond.cleanup: + ret void +} diff --git a/llvm/test/Analysis/CostModel/X86/interleaved-store-f64-stride-6.ll b/llvm/test/Analysis/CostModel/X86/interleaved-store-f64-stride-6.ll new file mode 100644 index 000000000000..8e1f9a7aa9c3 --- /dev/null +++ b/llvm/test/Analysis/CostModel/X86/interleaved-store-f64-stride-6.ll @@ -0,0 +1,84 @@ +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,SSE2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX1 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw,+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX512 +; REQUIRES: asserts + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@A = global [1024 x i8] zeroinitializer, align 128 +@B = global [1024 x double] zeroinitializer, align 128 + +; CHECK: LV: Checking a loop in "test" +; +; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store double %v5, double* %out5, align 8 +; SSE2: LV: Found an estimated cost of 20 for VF 2 For instruction: store double %v5, double* %out5, align 8 +; SSE2: LV: Found an estimated cost of 40 for VF 4 For instruction: store double %v5, double* %out5, align 8 +; +; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store double %v5, double* %out5, align 8 +; AVX1: LV: Found an estimated cost of 21 for VF 2 For instruction: store double %v5, double* %out5, align 8 +; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: store double %v5, double* %out5, align 8 +; AVX1: LV: Found an estimated cost of 108 for VF 8 For instruction: store double %v5, double* %out5, align 8 +; +; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store double %v5, double* %out5, align 8 +; AVX2: LV: Found an estimated cost of 21 for VF 2 For instruction: store double %v5, double* %out5, align 8 +; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: store double %v5, double* %out5, align 8 +; AVX2: LV: Found an estimated cost of 108 for VF 8 For instruction: store double %v5, double* %out5, align 8 +; +; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store double %v5, double* %out5, align 8 +; AVX512: LV: Found an estimated cost of 17 for VF 2 For instruction: store double %v5, double* %out5, align 8 +; AVX512: LV: Found an estimated cost of 25 for VF 4 For instruction: store double %v5, double* %out5, align 8 +; AVX512: LV: Found an estimated cost of 51 for VF 8 For instruction: store double %v5, double* %out5, align 8 +; AVX512: LV: Found an estimated cost of 102 for VF 16 For instruction: store double %v5, double* %out5, align 8 +; AVX512: LV: Found an estimated cost of 204 for VF 32 For instruction: store double %v5, double* %out5, align 8 +; +; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: store double %v5, double* %out5, align 8 + +define void @test() { +entry: + br label %for.body + +for.body: + %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ] + + %iv.0 = add nuw nsw i64 %iv, 0 + %iv.1 = add nuw nsw i64 %iv, 1 + %iv.2 = add nuw nsw i64 %iv, 2 + %iv.3 = add nuw nsw i64 %iv, 3 + %iv.4 = add nuw nsw i64 %iv, 4 + %iv.5 = add nuw nsw i64 %iv, 5 + + %in = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.0 + %v.narrow = load i8, i8* %in + + %v = uitofp i8 %v.narrow to double + + %v0 = fadd double %v, 0.0 + %v1 = fadd double %v, 1.0 + %v2 = fadd double %v, 2.0 + %v3 = fadd double %v, 3.0 + %v4 = fadd double %v, 4.0 + %v5 = fadd double %v, 5.0 + + %out0 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.0 + %out1 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.1 + %out2 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.2 + %out3 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.3 + %out4 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.4 + %out5 = getelementptr inbounds [1024 x double], [1024 x double]* @B, i64 0, i64 %iv.5 + + store double %v0, double* %out0 + store double %v1, double* %out1 + store double %v2, double* %out2 + store double %v3, double* %out3 + store double %v4, double* %out4 + store double %v5, double* %out5 + + %iv.next = add nuw nsw i64 %iv.0, 6 + %cmp = icmp ult i64 %iv.next, 1024 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.cond.cleanup: + ret void +} diff --git a/llvm/test/Analysis/CostModel/X86/interleaved-store-i64-stride-6.ll b/llvm/test/Analysis/CostModel/X86/interleaved-store-i64-stride-6.ll new file mode 100644 index 000000000000..1a1b9ca555fc --- /dev/null +++ b/llvm/test/Analysis/CostModel/X86/interleaved-store-i64-stride-6.ll @@ -0,0 +1,84 @@ +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,SSE2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX1 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX2 +; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw,+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX512 +; REQUIRES: asserts + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@A = global [1024 x i8] zeroinitializer, align 128 +@B = global [1024 x i64] zeroinitializer, align 128 + +; CHECK: LV: Checking a loop in "test" +; +; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i64 %v5, i64* %out5, align 8 +; SSE2: LV: Found an estimated cost of 44 for VF 2 For instruction: store i64 %v5, i64* %out5, align 8 +; SSE2: LV: Found an estimated cost of 88 for VF 4 For instruction: store i64 %v5, i64* %out5, align 8 +; +; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX1: LV: Found an estimated cost of 33 for VF 2 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX1: LV: Found an estimated cost of 78 for VF 4 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX1: LV: Found an estimated cost of 156 for VF 8 For instruction: store i64 %v5, i64* %out5, align 8 +; +; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX2: LV: Found an estimated cost of 33 for VF 2 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX2: LV: Found an estimated cost of 78 for VF 4 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX2: LV: Found an estimated cost of 156 for VF 8 For instruction: store i64 %v5, i64* %out5, align 8 +; +; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX512: LV: Found an estimated cost of 17 for VF 2 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX512: LV: Found an estimated cost of 25 for VF 4 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX512: LV: Found an estimated cost of 51 for VF 8 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX512: LV: Found an estimated cost of 102 for VF 16 For instruction: store i64 %v5, i64* %out5, align 8 +; AVX512: LV: Found an estimated cost of 204 for VF 32 For instruction: store i64 %v5, i64* %out5, align 8 +; +; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: store i64 %v5, i64* %out5, align 8 + +define void @test() { +entry: + br label %for.body + +for.body: + %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ] + + %iv.0 = add nuw nsw i64 %iv, 0 + %iv.1 = add nuw nsw i64 %iv, 1 + %iv.2 = add nuw nsw i64 %iv, 2 + %iv.3 = add nuw nsw i64 %iv, 3 + %iv.4 = add nuw nsw i64 %iv, 4 + %iv.5 = add nuw nsw i64 %iv, 5 + + %in = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.0 + %v.narrow = load i8, i8* %in + + %v = zext i8 %v.narrow to i64 + + %v0 = add i64 %v, 0 + %v1 = add i64 %v, 1 + %v2 = add i64 %v, 2 + %v3 = add i64 %v, 3 + %v4 = add i64 %v, 4 + %v5 = add i64 %v, 5 + + %out0 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.0 + %out1 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.1 + %out2 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.2 + %out3 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.3 + %out4 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.4 + %out5 = getelementptr inbounds [1024 x i64], [1024 x i64]* @B, i64 0, i64 %iv.5 + + store i64 %v0, i64* %out0 + store i64 %v1, i64* %out1 + store i64 %v2, i64* %out2 + store i64 %v3, i64* %out3 + store i64 %v4, i64* %out4 + store i64 %v5, i64* %out5 + + %iv.next = add nuw nsw i64 %iv.0, 6 + %cmp = icmp ult i64 %iv.next, 1024 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.cond.cleanup: + ret void +}