[llvm] 200edc1 - [NFC][X86][LV] Add basic costmodel test coverage for not-fully-interleaved i32 loads
Roman Lebedev via llvm-commits
llvm-commits at lists.llvm.org
Tue Oct 5 09:41:10 PDT 2021
Author: Roman Lebedev
Date: 2021-10-05T19:39:50+03:00
New Revision: 200edc152b029eb5e7eab56f73a0750935799035
URL: https://github.com/llvm/llvm-project/commit/200edc152b029eb5e7eab56f73a0750935799035
DIFF: https://github.com/llvm/llvm-project/commit/200edc152b029eb5e7eab56f73a0750935799035.diff
LOG: [NFC][X86][LV] Add basic costmodel test coverage for not-fully-interleaved i32 loads
The coverage could have cumulative explosion here,
so i'm adding only the most basic cases,
and hoping it's enough, though more can be added if needed.
Added:
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-2-indices-0u.ll
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-01u.ll
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-0uu.ll
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-012u.ll
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-01uu.ll
llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-0uuu.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-2-indices-0u.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-2-indices-0u.ll
new file mode 100644
index 000000000000..3080b0485e7f
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-2-indices-0u.ll
@@ -0,0 +1,71 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 236 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i32, i32* %in0, align 2
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+
+ %v0 = load i32, i32* %in0
+
+ %reduce.add.0 = add i32 %v0, 0
+
+ %reduce.add.0.narrow = trunc i32 %reduce.add.0 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.0.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 2
+ %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-i32-stride-3-indices-01u.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-01u.ll
new file mode 100644
index 000000000000..71b2992fcab9
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-01u.ll
@@ -0,0 +1,75 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 236 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i32, i32* %in0, align 4
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+ %in1 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.1
+
+ %v0 = load i32, i32* %in0
+ %v1 = load i32, i32* %in1
+
+ %reduce.add.0 = add i32 %v0, %v1
+ %reduce.add.1 = add i32 %reduce.add.0, 0
+
+ %reduce.add.1.narrow = trunc i32 %reduce.add.1 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.1.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 3
+ %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-i32-stride-3-indices-0uu.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-0uu.ll
new file mode 100644
index 000000000000..2ee62ea5bc62
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-3-indices-0uu.ll
@@ -0,0 +1,70 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 236 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+
+ %v0 = load i32, i32* %in0
+
+ %reduce.add.0 = add i32 %v0, 0
+ %reduce.add.1 = add i32 %reduce.add.0, 0
+
+ %reduce.add.1.narrow = trunc i32 %reduce.add.1 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.1.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 3
+ %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-i32-stride-4-indices-012u.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-012u.ll
new file mode 100644
index 000000000000..258a9901c2ea
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-012u.ll
@@ -0,0 +1,76 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i32, i32* %in0, align 4
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+ %in1 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.1
+ %in2 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.2
+
+ %v0 = load i32, i32* %in0
+ %v1 = load i32, i32* %in1
+ %v2 = load i32, i32* %in2
+
+ %reduce.add.0 = add i32 %v0, %v1
+ %reduce.add.1 = add i32 %reduce.add.0, %v2
+ %reduce.add.2 = add i32 %reduce.add.1, 0
+
+ %reduce.add.2.narrow = trunc i32 %reduce.add.2 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.2.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 4
+ %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-i32-stride-4-indices-01uu.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-01uu.ll
new file mode 100644
index 000000000000..dc191c308ff7
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-01uu.ll
@@ -0,0 +1,76 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 236 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i32, i32* %in0, align 4
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+ %in1 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.1
+
+ %v0 = load i32, i32* %in0
+ %v1 = load i32, i32* %in1
+
+ %reduce.add.0 = add i32 %v0, %v1
+ %reduce.add.1 = add i32 %reduce.add.0, 0
+ %reduce.add.2 = add i32 %reduce.add.1, 0
+
+ %reduce.add.2.narrow = trunc i32 %reduce.add.2 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.2.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 4
+ %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-i32-stride-4-indices-0uuu.ll b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-0uuu.ll
new file mode 100644
index 000000000000..0c7b21a4a299
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/interleaved-load-i32-stride-4-indices-0uuu.ll
@@ -0,0 +1,73 @@
+; 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"
+
+ at A = global [1024 x i32] zeroinitializer, align 128
+ at 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 i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 29 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 118 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; SSE2: LV: Found an estimated cost of 236 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX1: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 54 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 220 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX2: LV: Found an estimated cost of 440 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 6 for VF 4 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i32, i32* %in0, align 4
+; AVX512: LV: Found an estimated cost of 80 for VF 64 For instruction: %v0 = load i32, i32* %in0, align 4
+;
+; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i32, i32* %in0, align 4
+
+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
+
+ %in0 = getelementptr inbounds [1024 x i32], [1024 x i32]* @A, i64 0, i64 %iv.0
+
+ %v0 = load i32, i32* %in0
+
+ %reduce.add.0 = add i32 %v0, 0
+ %reduce.add.1 = add i32 %reduce.add.0, 0
+ %reduce.add.2 = add i32 %reduce.add.1, 0
+
+ %reduce.add.2.narrow = trunc i32 %reduce.add.2 to i8
+
+ %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
+ store i8 %reduce.add.2.narrow, i8* %out
+
+ %iv.next = add nuw nsw i64 %iv.0, 4
+ %cmp = icmp ult i64 %iv.next, 1024
+ br i1 %cmp, label %for.body, label %for.cond.cleanup
+
+for.cond.cleanup:
+ ret void
+}
More information about the llvm-commits
mailing list