summaryrefslogtreecommitdiff
path: root/test/Analysis/CostModel
diff options
context:
space:
mode:
Diffstat (limited to 'test/Analysis/CostModel')
-rw-r--r--test/Analysis/CostModel/SystemZ/div-pow2.ll154
-rw-r--r--test/Analysis/CostModel/X86/bitreverse.ll69
-rw-r--r--test/Analysis/CostModel/X86/ctbits-cost.ll587
-rw-r--r--test/Analysis/CostModel/X86/ctlz.ll233
-rw-r--r--test/Analysis/CostModel/X86/ctpop.ll133
-rw-r--r--test/Analysis/CostModel/X86/cttz.ll233
6 files changed, 822 insertions, 587 deletions
diff --git a/test/Analysis/CostModel/SystemZ/div-pow2.ll b/test/Analysis/CostModel/SystemZ/div-pow2.ll
new file mode 100644
index 0000000000000..9ef2dd71e8fa7
--- /dev/null
+++ b/test/Analysis/CostModel/SystemZ/div-pow2.ll
@@ -0,0 +1,154 @@
+; RUN: opt < %s -cost-model -analyze -mtriple=systemz-unknown -mcpu=z13 | FileCheck %s
+
+; Scalar sdiv
+
+define i64 @fun0(i64 %a) {
+ %r = sdiv i64 %a, 2
+ ret i64 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i64 %a, 2
+}
+
+define i64 @fun1(i64 %a) {
+ %r = sdiv i64 %a, -4
+ ret i64 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i64 %a, -4
+}
+
+define i32 @fun2(i32 %a) {
+ %r = sdiv i32 %a, 8
+ ret i32 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i32 %a, 8
+}
+
+define i32 @fun3(i32 %a) {
+ %r = sdiv i32 %a, -16
+ ret i32 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i32 %a, -16
+}
+
+define i16 @fun4(i16 %a) {
+ %r = sdiv i16 %a, 32
+ ret i16 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i16 %a, 32
+}
+
+define i16 @fun5(i16 %a) {
+ %r = sdiv i16 %a, -64
+ ret i16 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i16 %a, -64
+}
+
+define i8 @fun6(i8 %a) {
+ %r = sdiv i8 %a, 64
+ ret i8 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i8 %a, 64
+}
+
+define i8 @fun7(i8 %a) {
+ %r = sdiv i8 %a, -128
+ ret i8 %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv i8 %a, -128
+}
+
+
+; Vector sdiv
+
+define <2 x i64> @fun8(<2 x i64> %a) {
+ %r = sdiv <2 x i64> %a, <i64 2, i64 2>
+ ret <2 x i64> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <2 x i64> %a, <i64 2, i64 2>
+}
+
+define <2 x i64> @fun9(<2 x i64> %a) {
+ %r = sdiv <2 x i64> %a, <i64 -4, i64 -4>
+ ret <2 x i64> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <2 x i64> %a, <i64 -4, i64 -4>
+}
+
+define <4 x i32> @fun10(<4 x i32> %a) {
+ %r = sdiv <4 x i32> %a, <i32 8, i32 8, i32 8, i32 8>
+ ret <4 x i32> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <4 x i32> %a, <i32 8, i32 8, i32 8, i32 8>
+}
+
+define <4 x i32> @fun11(<4 x i32> %a) {
+ %r = sdiv <4 x i32> %a, <i32 -16, i32 -16, i32 -16, i32 -16>
+ ret <4 x i32> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <4 x i32> %a, <i32 -16
+}
+
+define <8 x i16> @fun12(<8 x i16> %a) {
+ %r = sdiv <8 x i16> %a, <i16 32, i16 32, i16 32, i16 32, i16 32, i16 32, i16 32, i16 32>
+ ret <8 x i16> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <8 x i16> %a, <i16 32
+}
+
+define <8 x i16> @fun13(<8 x i16> %a) {
+ %r = sdiv <8 x i16> %a, <i16 -64, i16 -64, i16 -64, i16 -64, i16 -64, i16 -64, i16 -64, i16 -64>
+ ret <8 x i16> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <8 x i16> %a, <i16 -64
+}
+
+define <16 x i8> @fun14(<16 x i8> %a) {
+ %r = sdiv <16 x i8> %a, <i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64, i8 64>
+ ret <16 x i8> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <16 x i8> %a, <i8 64
+}
+
+define <16 x i8> @fun15(<16 x i8> %a) {
+ %r = sdiv <16 x i8> %a, <i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128>
+ ret <16 x i8> %r
+; CHECK: Cost Model: Found an estimated cost of 4 for instruction: %r = sdiv <16 x i8> %a, <i8 -128
+}
+
+; Scalar udiv
+
+define i64 @fun16(i64 %a) {
+ %r = udiv i64 %a, 2
+ ret i64 %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv i64 %a, 2
+}
+
+define i32 @fun17(i32 %a) {
+ %r = udiv i32 %a, 8
+ ret i32 %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv i32 %a, 8
+}
+
+define i16 @fun18(i16 %a) {
+ %r = udiv i16 %a, 32
+ ret i16 %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv i16 %a, 32
+}
+
+define i8 @fun19(i8 %a) {
+ %r = udiv i8 %a, 128
+ ret i8 %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv i8 %a, -128
+}
+
+; Vector udiv
+
+define <2 x i64> @fun20(<2 x i64> %a) {
+ %r = udiv <2 x i64> %a, <i64 2, i64 2>
+ ret <2 x i64> %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv <2 x i64> %a, <i64 2
+}
+
+define <4 x i32> @fun21(<4 x i32> %a) {
+ %r = udiv <4 x i32> %a, <i32 8, i32 8, i32 8, i32 8>
+ ret <4 x i32> %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv <4 x i32> %a, <i32 8
+}
+
+define <8 x i16> @fun22(<8 x i16> %a) {
+ %r = udiv <8 x i16> %a, <i16 32, i16 32, i16 32, i16 32, i16 32, i16 32, i16 32, i16 32>
+ ret <8 x i16> %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv <8 x i16> %a, <i16 32
+}
+
+define <16 x i8> @fun23(<16 x i8> %a) {
+ %r = udiv <16 x i8> %a, <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>
+ ret <16 x i8> %r
+; CHECK: Cost Model: Found an estimated cost of 1 for instruction: %r = udiv <16 x i8> %a, <i8 -128
+}
diff --git a/test/Analysis/CostModel/X86/bitreverse.ll b/test/Analysis/CostModel/X86/bitreverse.ll
index 8d5e1421eb829..9321b7323b57f 100644
--- a/test/Analysis/CostModel/X86/bitreverse.ll
+++ b/test/Analysis/CostModel/X86/bitreverse.ll
@@ -2,10 +2,14 @@
; RUN: opt < %s -mtriple=i686-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X86 -check-prefix=SSE42
; RUN: opt < %s -mtriple=i686-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X86 -check-prefix=AVX
; RUN: opt < %s -mtriple=i686-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X86 -check-prefix=AVX2
+; RUN: opt < %s -mtriple=i686-unknown-linux-gnu -mcpu=knl -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X86 -check-prefix=AVX512 -check-prefix=AVX512F
+; RUN: opt < %s -mtriple=i686-unknown-linux-gnu -mcpu=skx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X86 -check-prefix=AVX512 -check-prefix=AVX512BW
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=pentium4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=SSE2
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=SSE42
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=AVX
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=knl -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=AVX512 -check-prefix=AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=skx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=X64 -check-prefix=AVX512 -check-prefix=AVX512BW
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=XOP -check-prefix=XOPAVX
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=XOP -check-prefix=XOPAVX2
@@ -64,12 +68,18 @@ declare <8 x i32> @llvm.bitreverse.v8i32(<8 x i32>)
declare <16 x i16> @llvm.bitreverse.v16i16(<16 x i16>)
declare <32 x i8> @llvm.bitreverse.v32i8(<32 x i8>)
+declare <8 x i64> @llvm.bitreverse.v8i64(<8 x i64>)
+declare <16 x i32> @llvm.bitreverse.v16i32(<16 x i32>)
+declare <32 x i16> @llvm.bitreverse.v32i16(<32 x i16>)
+declare <64 x i8> @llvm.bitreverse.v64i8(<64 x i8>)
+
define <2 x i64> @var_bitreverse_v2i64(<2 x i64> %a) {
; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v2i64':
; SSE2: Found an estimated cost of 29 for instruction: %bitreverse
; SSE42: Found an estimated cost of 5 for instruction: %bitreverse
; AVX: Found an estimated cost of 5 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 1 for instruction: %bitreverse
%bitreverse = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> %a)
ret <2 x i64> %bitreverse
@@ -81,17 +91,32 @@ define <4 x i64> @var_bitreverse_v4i64(<4 x i64> %a) {
; SSE42: Found an estimated cost of 10 for instruction: %bitreverse
; AVX: Found an estimated cost of 12 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 4 for instruction: %bitreverse
%bitreverse = call <4 x i64> @llvm.bitreverse.v4i64(<4 x i64> %a)
ret <4 x i64> %bitreverse
}
+define <8 x i64> @var_bitreverse_v8i64(<8 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v8i64':
+; SSE2: Found an estimated cost of 116 for instruction: %bitreverse
+; SSE42: Found an estimated cost of 20 for instruction: %bitreverse
+; AVX: Found an estimated cost of 24 for instruction: %bitreverse
+; AVX2: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512F: Found an estimated cost of 36 for instruction: %bitreverse
+; AVX512BW: Found an estimated cost of 5 for instruction: %bitreverse
+; XOP: Found an estimated cost of 8 for instruction: %bitreverse
+ %bitreverse = call <8 x i64> @llvm.bitreverse.v8i64(<8 x i64> %a)
+ ret <8 x i64> %bitreverse
+}
+
define <4 x i32> @var_bitreverse_v4i32(<4 x i32> %a) {
; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v4i32':
; SSE2: Found an estimated cost of 27 for instruction: %bitreverse
; SSE42: Found an estimated cost of 5 for instruction: %bitreverse
; AVX: Found an estimated cost of 5 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 1 for instruction: %bitreverse
%bitreverse = call <4 x i32> @llvm.bitreverse.v4i32(<4 x i32> %a)
ret <4 x i32> %bitreverse
@@ -103,17 +128,32 @@ define <8 x i32> @var_bitreverse_v8i32(<8 x i32> %a) {
; SSE42: Found an estimated cost of 10 for instruction: %bitreverse
; AVX: Found an estimated cost of 12 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 4 for instruction: %bitreverse
%bitreverse = call <8 x i32> @llvm.bitreverse.v8i32(<8 x i32> %a)
ret <8 x i32> %bitreverse
}
+define <16 x i32> @var_bitreverse_v16i32(<16 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v16i32':
+; SSE2: Found an estimated cost of 108 for instruction: %bitreverse
+; SSE42: Found an estimated cost of 20 for instruction: %bitreverse
+; AVX: Found an estimated cost of 24 for instruction: %bitreverse
+; AVX2: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512F: Found an estimated cost of 24 for instruction: %bitreverse
+; AVX512BW: Found an estimated cost of 5 for instruction: %bitreverse
+; XOP: Found an estimated cost of 8 for instruction: %bitreverse
+ %bitreverse = call <16 x i32> @llvm.bitreverse.v16i32(<16 x i32> %a)
+ ret <16 x i32> %bitreverse
+}
+
define <8 x i16> @var_bitreverse_v8i16(<8 x i16> %a) {
; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v8i16':
; SSE2: Found an estimated cost of 27 for instruction: %bitreverse
; SSE42: Found an estimated cost of 5 for instruction: %bitreverse
; AVX: Found an estimated cost of 5 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 1 for instruction: %bitreverse
%bitreverse = call <8 x i16> @llvm.bitreverse.v8i16(<8 x i16> %a)
ret <8 x i16> %bitreverse
@@ -125,17 +165,32 @@ define <16 x i16> @var_bitreverse_v16i16(<16 x i16> %a) {
; SSE42: Found an estimated cost of 10 for instruction: %bitreverse
; AVX: Found an estimated cost of 12 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 4 for instruction: %bitreverse
%bitreverse = call <16 x i16> @llvm.bitreverse.v16i16(<16 x i16> %a)
ret <16 x i16> %bitreverse
}
+define <32 x i16> @var_bitreverse_v32i16(<32 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v32i16':
+; SSE2: Found an estimated cost of 108 for instruction: %bitreverse
+; SSE42: Found an estimated cost of 20 for instruction: %bitreverse
+; AVX: Found an estimated cost of 24 for instruction: %bitreverse
+; AVX2: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512F: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512BW: Found an estimated cost of 5 for instruction: %bitreverse
+; XOP: Found an estimated cost of 8 for instruction: %bitreverse
+ %bitreverse = call <32 x i16> @llvm.bitreverse.v32i16(<32 x i16> %a)
+ ret <32 x i16> %bitreverse
+}
+
define <16 x i8> @var_bitreverse_v16i8(<16 x i8> %a) {
; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v16i8':
; SSE2: Found an estimated cost of 20 for instruction: %bitreverse
; SSE42: Found an estimated cost of 5 for instruction: %bitreverse
; AVX: Found an estimated cost of 5 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 1 for instruction: %bitreverse
%bitreverse = call <16 x i8> @llvm.bitreverse.v16i8(<16 x i8> %a)
ret <16 x i8> %bitreverse
@@ -147,7 +202,21 @@ define <32 x i8> @var_bitreverse_v32i8(<32 x i8> %a) {
; SSE42: Found an estimated cost of 10 for instruction: %bitreverse
; AVX: Found an estimated cost of 12 for instruction: %bitreverse
; AVX2: Found an estimated cost of 5 for instruction: %bitreverse
+; AVX512: Found an estimated cost of 5 for instruction: %bitreverse
; XOP: Found an estimated cost of 4 for instruction: %bitreverse
%bitreverse = call <32 x i8> @llvm.bitreverse.v32i8(<32 x i8> %a)
ret <32 x i8> %bitreverse
}
+
+define <64 x i8> @var_bitreverse_v64i8(<64 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_bitreverse_v64i8':
+; SSE2: Found an estimated cost of 80 for instruction: %bitreverse
+; SSE42: Found an estimated cost of 20 for instruction: %bitreverse
+; AVX: Found an estimated cost of 24 for instruction: %bitreverse
+; AVX2: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512F: Found an estimated cost of 10 for instruction: %bitreverse
+; AVX512BW: Found an estimated cost of 5 for instruction: %bitreverse
+; XOP: Found an estimated cost of 8 for instruction: %bitreverse
+ %bitreverse = call <64 x i8> @llvm.bitreverse.v64i8(<64 x i8> %a)
+ ret <64 x i8> %bitreverse
+}
diff --git a/test/Analysis/CostModel/X86/ctbits-cost.ll b/test/Analysis/CostModel/X86/ctbits-cost.ll
deleted file mode 100644
index aaf092c7b1d75..0000000000000
--- a/test/Analysis/CostModel/X86/ctbits-cost.ll
+++ /dev/null
@@ -1,587 +0,0 @@
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=pentium4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE2 -check-prefix=NOPOPCNT
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE42 -check-prefix=POPCNT
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
-
-; Verify the cost of scalar population count instructions.
-
-declare i64 @llvm.ctpop.i64(i64)
-declare i32 @llvm.ctpop.i32(i32)
-declare i16 @llvm.ctpop.i16(i16)
-declare i8 @llvm.ctpop.i8(i8)
-
-define i64 @var_ctpop_i64(i64 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i64':
-; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
-; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
- %ctpop = call i64 @llvm.ctpop.i64(i64 %a)
- ret i64 %ctpop
-}
-
-define i32 @var_ctpop_i32(i32 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i32':
-; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
-; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
- %ctpop = call i32 @llvm.ctpop.i32(i32 %a)
- ret i32 %ctpop
-}
-
-define i16 @var_ctpop_i16(i16 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i16':
-; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
-; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
- %ctpop = call i16 @llvm.ctpop.i16(i16 %a)
- ret i16 %ctpop
-}
-
-define i8 @var_ctpop_i8(i8 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i8':
-; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
-; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
- %ctpop = call i8 @llvm.ctpop.i8(i8 %a)
- ret i8 %ctpop
-}
-
-; Verify the cost of vector population count instructions.
-
-declare <2 x i64> @llvm.ctpop.v2i64(<2 x i64>)
-declare <4 x i32> @llvm.ctpop.v4i32(<4 x i32>)
-declare <8 x i16> @llvm.ctpop.v8i16(<8 x i16>)
-declare <16 x i8> @llvm.ctpop.v16i8(<16 x i8>)
-
-declare <4 x i64> @llvm.ctpop.v4i64(<4 x i64>)
-declare <8 x i32> @llvm.ctpop.v8i32(<8 x i32>)
-declare <16 x i16> @llvm.ctpop.v16i16(<16 x i16>)
-declare <32 x i8> @llvm.ctpop.v32i8(<32 x i8>)
-
-define <2 x i64> @var_ctpop_v2i64(<2 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v2i64':
-; SSE2: Found an estimated cost of 12 for instruction: %ctpop
-; SSE42: Found an estimated cost of 7 for instruction: %ctpop
-; AVX: Found an estimated cost of 7 for instruction: %ctpop
- %ctpop = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a)
- ret <2 x i64> %ctpop
-}
-
-define <4 x i64> @var_ctpop_v4i64(<4 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v4i64':
-; SSE2: Found an estimated cost of 24 for instruction: %ctpop
-; SSE42: Found an estimated cost of 14 for instruction: %ctpop
-; AVX1: Found an estimated cost of 16 for instruction: %ctpop
-; AVX2: Found an estimated cost of 7 for instruction: %ctpop
- %ctpop = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> %a)
- ret <4 x i64> %ctpop
-}
-
-define <4 x i32> @var_ctpop_v4i32(<4 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v4i32':
-; SSE2: Found an estimated cost of 15 for instruction: %ctpop
-; SSE42: Found an estimated cost of 11 for instruction: %ctpop
-; AVX: Found an estimated cost of 11 for instruction: %ctpop
- %ctpop = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> %a)
- ret <4 x i32> %ctpop
-}
-
-define <8 x i32> @var_ctpop_v8i32(<8 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v8i32':
-; SSE2: Found an estimated cost of 30 for instruction: %ctpop
-; SSE42: Found an estimated cost of 22 for instruction: %ctpop
-; AVX1: Found an estimated cost of 24 for instruction: %ctpop
-; AVX2: Found an estimated cost of 11 for instruction: %ctpop
- %ctpop = call <8 x i32> @llvm.ctpop.v8i32(<8 x i32> %a)
- ret <8 x i32> %ctpop
-}
-
-define <8 x i16> @var_ctpop_v8i16(<8 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v8i16':
-; SSE2: Found an estimated cost of 13 for instruction: %ctpop
-; SSE42: Found an estimated cost of 9 for instruction: %ctpop
-; AVX: Found an estimated cost of 9 for instruction: %ctpop
- %ctpop = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> %a)
- ret <8 x i16> %ctpop
-}
-
-define <16 x i16> @var_ctpop_v16i16(<16 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v16i16':
-; SSE2: Found an estimated cost of 26 for instruction: %ctpop
-; SSE42: Found an estimated cost of 18 for instruction: %ctpop
-; AVX1: Found an estimated cost of 20 for instruction: %ctpop
-; AVX2: Found an estimated cost of 9 for instruction: %ctpop
- %ctpop = call <16 x i16> @llvm.ctpop.v16i16(<16 x i16> %a)
- ret <16 x i16> %ctpop
-}
-
-define <16 x i8> @var_ctpop_v16i8(<16 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v16i8':
-; SSE2: Found an estimated cost of 10 for instruction: %ctpop
-; SSE42: Found an estimated cost of 6 for instruction: %ctpop
-; AVX: Found an estimated cost of 6 for instruction: %ctpop
- %ctpop = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> %a)
- ret <16 x i8> %ctpop
-}
-
-define <32 x i8> @var_ctpop_v32i8(<32 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v32i8':
-; SSE2: Found an estimated cost of 20 for instruction: %ctpop
-; SSE42: Found an estimated cost of 12 for instruction: %ctpop
-; AVX1: Found an estimated cost of 14 for instruction: %ctpop
-; AVX2: Found an estimated cost of 6 for instruction: %ctpop
- %ctpop = call <32 x i8> @llvm.ctpop.v32i8(<32 x i8> %a)
- ret <32 x i8> %ctpop
-}
-
-; Verify the cost of scalar leading zero count instructions.
-
-declare i64 @llvm.ctlz.i64(i64, i1)
-declare i32 @llvm.ctlz.i32(i32, i1)
-declare i16 @llvm.ctlz.i16(i16, i1)
-declare i8 @llvm.ctlz.i8(i8, i1)
-
-define i64 @var_ctlz_i64(i64 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i64':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i64 @llvm.ctlz.i64(i64 %a, i1 0)
- ret i64 %ctlz
-}
-
-define i64 @var_ctlz_i64u(i64 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i64u':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i64 @llvm.ctlz.i64(i64 %a, i1 1)
- ret i64 %ctlz
-}
-
-define i32 @var_ctlz_i32(i32 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i32':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i32 @llvm.ctlz.i32(i32 %a, i1 0)
- ret i32 %ctlz
-}
-
-define i32 @var_ctlz_i32u(i32 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i32u':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i32 @llvm.ctlz.i32(i32 %a, i1 1)
- ret i32 %ctlz
-}
-
-define i16 @var_ctlz_i16(i16 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i16':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i16 @llvm.ctlz.i16(i16 %a, i1 0)
- ret i16 %ctlz
-}
-
-define i16 @var_ctlz_i16u(i16 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i16u':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i16 @llvm.ctlz.i16(i16 %a, i1 1)
- ret i16 %ctlz
-}
-
-define i8 @var_ctlz_i8(i8 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i8':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i8 @llvm.ctlz.i8(i8 %a, i1 0)
- ret i8 %ctlz
-}
-
-define i8 @var_ctlz_i8u(i8 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i8u':
-; CHECK: Found an estimated cost of 1 for instruction: %ctlz
- %ctlz = call i8 @llvm.ctlz.i8(i8 %a, i1 1)
- ret i8 %ctlz
-}
-
-; Verify the cost of vector leading zero count instructions.
-
-declare <2 x i64> @llvm.ctlz.v2i64(<2 x i64>, i1)
-declare <4 x i32> @llvm.ctlz.v4i32(<4 x i32>, i1)
-declare <8 x i16> @llvm.ctlz.v8i16(<8 x i16>, i1)
-declare <16 x i8> @llvm.ctlz.v16i8(<16 x i8>, i1)
-
-declare <4 x i64> @llvm.ctlz.v4i64(<4 x i64>, i1)
-declare <8 x i32> @llvm.ctlz.v8i32(<8 x i32>, i1)
-declare <16 x i16> @llvm.ctlz.v16i16(<16 x i16>, i1)
-declare <32 x i8> @llvm.ctlz.v32i8(<32 x i8>, i1)
-
-define <2 x i64> @var_ctlz_v2i64(<2 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v2i64':
-; SSE2: Found an estimated cost of 25 for instruction: %ctlz
-; SSE42: Found an estimated cost of 23 for instruction: %ctlz
-; AVX: Found an estimated cost of 23 for instruction: %ctlz
- %ctlz = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 0)
- ret <2 x i64> %ctlz
-}
-
-define <2 x i64> @var_ctlz_v2i64u(<2 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v2i64u':
-; SSE2: Found an estimated cost of 25 for instruction: %ctlz
-; SSE42: Found an estimated cost of 23 for instruction: %ctlz
-; AVX: Found an estimated cost of 23 for instruction: %ctlz
- %ctlz = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 1)
- ret <2 x i64> %ctlz
-}
-
-define <4 x i64> @var_ctlz_v4i64(<4 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i64':
-; SSE2: Found an estimated cost of 50 for instruction: %ctlz
-; SSE42: Found an estimated cost of 46 for instruction: %ctlz
-; AVX1: Found an estimated cost of 48 for instruction: %ctlz
-; AVX2: Found an estimated cost of 23 for instruction: %ctlz
- %ctlz = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> %a, i1 0)
- ret <4 x i64> %ctlz
-}
-
-define <4 x i64> @var_ctlz_v4i64u(<4 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i64u':
-; SSE2: Found an estimated cost of 50 for instruction: %ctlz
-; SSE42: Found an estimated cost of 46 for instruction: %ctlz
-; AVX1: Found an estimated cost of 48 for instruction: %ctlz
-; AVX2: Found an estimated cost of 23 for instruction: %ctlz
- %ctlz = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> %a, i1 1)
- ret <4 x i64> %ctlz
-}
-
-define <4 x i32> @var_ctlz_v4i32(<4 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i32':
-; SSE2: Found an estimated cost of 26 for instruction: %ctlz
-; SSE42: Found an estimated cost of 18 for instruction: %ctlz
-; AVX: Found an estimated cost of 18 for instruction: %ctlz
- %ctlz = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %a, i1 0)
- ret <4 x i32> %ctlz
-}
-
-define <4 x i32> @var_ctlz_v4i32u(<4 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i32u':
-; SSE2: Found an estimated cost of 26 for instruction: %ctlz
-; SSE42: Found an estimated cost of 18 for instruction: %ctlz
-; AVX: Found an estimated cost of 18 for instruction: %ctlz
- %ctlz = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %a, i1 1)
- ret <4 x i32> %ctlz
-}
-
-define <8 x i32> @var_ctlz_v8i32(<8 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i32':
-; SSE2: Found an estimated cost of 52 for instruction: %ctlz
-; SSE42: Found an estimated cost of 36 for instruction: %ctlz
-; AVX1: Found an estimated cost of 38 for instruction: %ctlz
-; AVX2: Found an estimated cost of 18 for instruction: %ctlz
- %ctlz = call <8 x i32> @llvm.ctlz.v8i32(<8 x i32> %a, i1 0)
- ret <8 x i32> %ctlz
-}
-
-define <8 x i32> @var_ctlz_v8i32u(<8 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i32u':
-; SSE2: Found an estimated cost of 52 for instruction: %ctlz
-; SSE42: Found an estimated cost of 36 for instruction: %ctlz
-; AVX1: Found an estimated cost of 38 for instruction: %ctlz
-; AVX2: Found an estimated cost of 18 for instruction: %ctlz
- %ctlz = call <8 x i32> @llvm.ctlz.v8i32(<8 x i32> %a, i1 1)
- ret <8 x i32> %ctlz
-}
-
-define <8 x i16> @var_ctlz_v8i16(<8 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i16':
-; SSE2: Found an estimated cost of 20 for instruction: %ctlz
-; SSE42: Found an estimated cost of 14 for instruction: %ctlz
-; AVX: Found an estimated cost of 14 for instruction: %ctlz
- %ctlz = call <8 x i16> @llvm.ctlz.v8i16(<8 x i16> %a, i1 0)
- ret <8 x i16> %ctlz
-}
-
-define <8 x i16> @var_ctlz_v8i16u(<8 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i16u':
-; SSE2: Found an estimated cost of 20 for instruction: %ctlz
-; SSE42: Found an estimated cost of 14 for instruction: %ctlz
-; AVX: Found an estimated cost of 14 for instruction: %ctlz
- %ctlz = call <8 x i16> @llvm.ctlz.v8i16(<8 x i16> %a, i1 1)
- ret <8 x i16> %ctlz
-}
-
-define <16 x i16> @var_ctlz_v16i16(<16 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i16':
-; SSE2: Found an estimated cost of 40 for instruction: %ctlz
-; SSE42: Found an estimated cost of 28 for instruction: %ctlz
-; AVX1: Found an estimated cost of 30 for instruction: %ctlz
-; AVX2: Found an estimated cost of 14 for instruction: %ctlz
- %ctlz = call <16 x i16> @llvm.ctlz.v16i16(<16 x i16> %a, i1 0)
- ret <16 x i16> %ctlz
-}
-
-define <16 x i16> @var_ctlz_v16i16u(<16 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i16u':
-; SSE2: Found an estimated cost of 40 for instruction: %ctlz
-; SSE42: Found an estimated cost of 28 for instruction: %ctlz
-; AVX1: Found an estimated cost of 30 for instruction: %ctlz
-; AVX2: Found an estimated cost of 14 for instruction: %ctlz
- %ctlz = call <16 x i16> @llvm.ctlz.v16i16(<16 x i16> %a, i1 1)
- ret <16 x i16> %ctlz
-}
-
-define <16 x i8> @var_ctlz_v16i8(<16 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i8':
-; SSE2: Found an estimated cost of 17 for instruction: %ctlz
-; SSE42: Found an estimated cost of 9 for instruction: %ctlz
-; AVX: Found an estimated cost of 9 for instruction: %ctlz
- %ctlz = call <16 x i8> @llvm.ctlz.v16i8(<16 x i8> %a, i1 0)
- ret <16 x i8> %ctlz
-}
-
-define <16 x i8> @var_ctlz_v16i8u(<16 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i8u':
-; SSE2: Found an estimated cost of 17 for instruction: %ctlz
-; SSE42: Found an estimated cost of 9 for instruction: %ctlz
-; AVX: Found an estimated cost of 9 for instruction: %ctlz
- %ctlz = call <16 x i8> @llvm.ctlz.v16i8(<16 x i8> %a, i1 1)
- ret <16 x i8> %ctlz
-}
-
-define <32 x i8> @var_ctlz_v32i8(<32 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v32i8':
-; SSE2: Found an estimated cost of 34 for instruction: %ctlz
-; SSE42: Found an estimated cost of 18 for instruction: %ctlz
-; AVX1: Found an estimated cost of 20 for instruction: %ctlz
-; AVX2: Found an estimated cost of 9 for instruction: %ctlz
- %ctlz = call <32 x i8> @llvm.ctlz.v32i8(<32 x i8> %a, i1 0)
- ret <32 x i8> %ctlz
-}
-
-define <32 x i8> @var_ctlz_v32i8u(<32 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v32i8u':
-; SSE2: Found an estimated cost of 34 for instruction: %ctlz
-; SSE42: Found an estimated cost of 18 for instruction: %ctlz
-; AVX1: Found an estimated cost of 20 for instruction: %ctlz
-; AVX2: Found an estimated cost of 9 for instruction: %ctlz
- %ctlz = call <32 x i8> @llvm.ctlz.v32i8(<32 x i8> %a, i1 1)
- ret <32 x i8> %ctlz
-}
-
-; Verify the cost of scalar trailing zero count instructions.
-
-declare i64 @llvm.cttz.i64(i64, i1)
-declare i32 @llvm.cttz.i32(i32, i1)
-declare i16 @llvm.cttz.i16(i16, i1)
-declare i8 @llvm.cttz.i8(i8, i1)
-
-define i64 @var_cttz_i64(i64 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i64':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i64 @llvm.cttz.i64(i64 %a, i1 0)
- ret i64 %cttz
-}
-
-define i64 @var_cttz_i64u(i64 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i64u':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i64 @llvm.cttz.i64(i64 %a, i1 1)
- ret i64 %cttz
-}
-
-define i32 @var_cttz_i32(i32 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i32':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i32 @llvm.cttz.i32(i32 %a, i1 0)
- ret i32 %cttz
-}
-
-define i32 @var_cttz_i32u(i32 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i32u':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i32 @llvm.cttz.i32(i32 %a, i1 1)
- ret i32 %cttz
-}
-
-define i16 @var_cttz_i16(i16 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i16':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i16 @llvm.cttz.i16(i16 %a, i1 0)
- ret i16 %cttz
-}
-
-define i16 @var_cttz_i16u(i16 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i16u':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i16 @llvm.cttz.i16(i16 %a, i1 1)
- ret i16 %cttz
-}
-
-define i8 @var_cttz_i8(i8 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i8':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i8 @llvm.cttz.i8(i8 %a, i1 0)
- ret i8 %cttz
-}
-
-define i8 @var_cttz_i8u(i8 %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_i8u':
-; CHECK: Found an estimated cost of 1 for instruction: %cttz
- %cttz = call i8 @llvm.cttz.i8(i8 %a, i1 1)
- ret i8 %cttz
-}
-
-; Verify the cost of vector trailing zero count instructions.
-
-declare <2 x i64> @llvm.cttz.v2i64(<2 x i64>, i1)
-declare <4 x i32> @llvm.cttz.v4i32(<4 x i32>, i1)
-declare <8 x i16> @llvm.cttz.v8i16(<8 x i16>, i1)
-declare <16 x i8> @llvm.cttz.v16i8(<16 x i8>, i1)
-
-declare <4 x i64> @llvm.cttz.v4i64(<4 x i64>, i1)
-declare <8 x i32> @llvm.cttz.v8i32(<8 x i32>, i1)
-declare <16 x i16> @llvm.cttz.v16i16(<16 x i16>, i1)
-declare <32 x i8> @llvm.cttz.v32i8(<32 x i8>, i1)
-
-define <2 x i64> @var_cttz_v2i64(<2 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v2i64':
-; SSE2: Found an estimated cost of 14 for instruction: %cttz
-; SSE42: Found an estimated cost of 10 for instruction: %cttz
-; AVX: Found an estimated cost of 10 for instruction: %cttz
- %cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 0)
- ret <2 x i64> %cttz
-}
-
-define <2 x i64> @var_cttz_v2i64u(<2 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v2i64u':
-; SSE2: Found an estimated cost of 14 for instruction: %cttz
-; SSE42: Found an estimated cost of 10 for instruction: %cttz
-; AVX: Found an estimated cost of 10 for instruction: %cttz
- %cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 1)
- ret <2 x i64> %cttz
-}
-
-define <4 x i64> @var_cttz_v4i64(<4 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i64':
-; SSE2: Found an estimated cost of 28 for instruction: %cttz
-; SSE42: Found an estimated cost of 20 for instruction: %cttz
-; AVX1: Found an estimated cost of 22 for instruction: %cttz
-; AVX2: Found an estimated cost of 10 for instruction: %cttz
- %cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 0)
- ret <4 x i64> %cttz
-}
-
-define <4 x i64> @var_cttz_v4i64u(<4 x i64> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i64u':
-; SSE2: Found an estimated cost of 28 for instruction: %cttz
-; SSE42: Found an estimated cost of 20 for instruction: %cttz
-; AVX1: Found an estimated cost of 22 for instruction: %cttz
-; AVX2: Found an estimated cost of 10 for instruction: %cttz
- %cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 1)
- ret <4 x i64> %cttz
-}
-
-define <4 x i32> @var_cttz_v4i32(<4 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i32':
-; SSE2: Found an estimated cost of 18 for instruction: %cttz
-; SSE42: Found an estimated cost of 14 for instruction: %cttz
-; AVX: Found an estimated cost of 14 for instruction: %cttz
- %cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 0)
- ret <4 x i32> %cttz
-}
-
-define <4 x i32> @var_cttz_v4i32u(<4 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i32u':
-; SSE2: Found an estimated cost of 18 for instruction: %cttz
-; SSE42: Found an estimated cost of 14 for instruction: %cttz
-; AVX: Found an estimated cost of 14 for instruction: %cttz
- %cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 1)
- ret <4 x i32> %cttz
-}
-
-define <8 x i32> @var_cttz_v8i32(<8 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i32':
-; SSE2: Found an estimated cost of 36 for instruction: %cttz
-; SSE42: Found an estimated cost of 28 for instruction: %cttz
-; AVX1: Found an estimated cost of 30 for instruction: %cttz
-; AVX2: Found an estimated cost of 14 for instruction: %cttz
- %cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 0)
- ret <8 x i32> %cttz
-}
-
-define <8 x i32> @var_cttz_v8i32u(<8 x i32> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i32u':
-; SSE2: Found an estimated cost of 36 for instruction: %cttz
-; SSE42: Found an estimated cost of 28 for instruction: %cttz
-; AVX1: Found an estimated cost of 30 for instruction: %cttz
-; AVX2: Found an estimated cost of 14 for instruction: %cttz
- %cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 1)
- ret <8 x i32> %cttz
-}
-
-define <8 x i16> @var_cttz_v8i16(<8 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i16':
-; SSE2: Found an estimated cost of 16 for instruction: %cttz
-; SSE42: Found an estimated cost of 12 for instruction: %cttz
-; AVX: Found an estimated cost of 12 for instruction: %cttz
- %cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 0)
- ret <8 x i16> %cttz
-}
-
-define <8 x i16> @var_cttz_v8i16u(<8 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i16u':
-; SSE2: Found an estimated cost of 16 for instruction: %cttz
-; SSE42: Found an estimated cost of 12 for instruction: %cttz
-; AVX: Found an estimated cost of 12 for instruction: %cttz
- %cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 1)
- ret <8 x i16> %cttz
-}
-
-define <16 x i16> @var_cttz_v16i16(<16 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i16':
-; SSE2: Found an estimated cost of 32 for instruction: %cttz
-; SSE42: Found an estimated cost of 24 for instruction: %cttz
-; AVX1: Found an estimated cost of 26 for instruction: %cttz
-; AVX2: Found an estimated cost of 12 for instruction: %cttz
- %cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 0)
- ret <16 x i16> %cttz
-}
-
-define <16 x i16> @var_cttz_v16i16u(<16 x i16> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i16u':
-; SSE2: Found an estimated cost of 32 for instruction: %cttz
-; SSE42: Found an estimated cost of 24 for instruction: %cttz
-; AVX1: Found an estimated cost of 26 for instruction: %cttz
-; AVX2: Found an estimated cost of 12 for instruction: %cttz
- %cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 1)
- ret <16 x i16> %cttz
-}
-
-define <16 x i8> @var_cttz_v16i8(<16 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i8':
-; SSE2: Found an estimated cost of 13 for instruction: %cttz
-; SSE42: Found an estimated cost of 9 for instruction: %cttz
-; AVX: Found an estimated cost of 9 for instruction: %cttz
- %cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 0)
- ret <16 x i8> %cttz
-}
-
-define <16 x i8> @var_cttz_v16i8u(<16 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i8u':
-; SSE2: Found an estimated cost of 13 for instruction: %cttz
-; SSE42: Found an estimated cost of 9 for instruction: %cttz
-; AVX: Found an estimated cost of 9 for instruction: %cttz
- %cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 1)
- ret <16 x i8> %cttz
-}
-
-define <32 x i8> @var_cttz_v32i8(<32 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v32i8':
-; SSE2: Found an estimated cost of 26 for instruction: %cttz
-; SSE42: Found an estimated cost of 18 for instruction: %cttz
-; AVX1: Found an estimated cost of 20 for instruction: %cttz
-; AVX2: Found an estimated cost of 9 for instruction: %cttz
- %cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 0)
- ret <32 x i8> %cttz
-}
-
-define <32 x i8> @var_cttz_v32i8u(<32 x i8> %a) {
-; CHECK: 'Cost Model Analysis' for function 'var_cttz_v32i8u':
-; SSE2: Found an estimated cost of 26 for instruction: %cttz
-; SSE42: Found an estimated cost of 18 for instruction: %cttz
-; AVX1: Found an estimated cost of 20 for instruction: %cttz
-; AVX2: Found an estimated cost of 9 for instruction: %cttz
- %cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 1)
- ret <32 x i8> %cttz
-}
diff --git a/test/Analysis/CostModel/X86/ctlz.ll b/test/Analysis/CostModel/X86/ctlz.ll
new file mode 100644
index 0000000000000..2c97da15aee5b
--- /dev/null
+++ b/test/Analysis/CostModel/X86/ctlz.ll
@@ -0,0 +1,233 @@
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=pentium4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE2 -check-prefix=NOPOPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE42 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+
+; Verify the cost of scalar leading zero count instructions.
+
+declare i64 @llvm.ctlz.i64(i64, i1)
+declare i32 @llvm.ctlz.i32(i32, i1)
+declare i16 @llvm.ctlz.i16(i16, i1)
+declare i8 @llvm.ctlz.i8(i8, i1)
+
+define i64 @var_ctlz_i64(i64 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i64':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i64 @llvm.ctlz.i64(i64 %a, i1 0)
+ ret i64 %ctlz
+}
+
+define i64 @var_ctlz_i64u(i64 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i64u':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i64 @llvm.ctlz.i64(i64 %a, i1 1)
+ ret i64 %ctlz
+}
+
+define i32 @var_ctlz_i32(i32 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i32':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i32 @llvm.ctlz.i32(i32 %a, i1 0)
+ ret i32 %ctlz
+}
+
+define i32 @var_ctlz_i32u(i32 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i32u':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i32 @llvm.ctlz.i32(i32 %a, i1 1)
+ ret i32 %ctlz
+}
+
+define i16 @var_ctlz_i16(i16 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i16':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i16 @llvm.ctlz.i16(i16 %a, i1 0)
+ ret i16 %ctlz
+}
+
+define i16 @var_ctlz_i16u(i16 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i16u':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i16 @llvm.ctlz.i16(i16 %a, i1 1)
+ ret i16 %ctlz
+}
+
+define i8 @var_ctlz_i8(i8 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i8':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i8 @llvm.ctlz.i8(i8 %a, i1 0)
+ ret i8 %ctlz
+}
+
+define i8 @var_ctlz_i8u(i8 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_i8u':
+; CHECK: Found an estimated cost of 1 for instruction: %ctlz
+ %ctlz = call i8 @llvm.ctlz.i8(i8 %a, i1 1)
+ ret i8 %ctlz
+}
+
+; Verify the cost of vector leading zero count instructions.
+
+declare <2 x i64> @llvm.ctlz.v2i64(<2 x i64>, i1)
+declare <4 x i32> @llvm.ctlz.v4i32(<4 x i32>, i1)
+declare <8 x i16> @llvm.ctlz.v8i16(<8 x i16>, i1)
+declare <16 x i8> @llvm.ctlz.v16i8(<16 x i8>, i1)
+
+declare <4 x i64> @llvm.ctlz.v4i64(<4 x i64>, i1)
+declare <8 x i32> @llvm.ctlz.v8i32(<8 x i32>, i1)
+declare <16 x i16> @llvm.ctlz.v16i16(<16 x i16>, i1)
+declare <32 x i8> @llvm.ctlz.v32i8(<32 x i8>, i1)
+
+define <2 x i64> @var_ctlz_v2i64(<2 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v2i64':
+; SSE2: Found an estimated cost of 25 for instruction: %ctlz
+; SSE42: Found an estimated cost of 23 for instruction: %ctlz
+; AVX: Found an estimated cost of 23 for instruction: %ctlz
+ %ctlz = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 0)
+ ret <2 x i64> %ctlz
+}
+
+define <2 x i64> @var_ctlz_v2i64u(<2 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v2i64u':
+; SSE2: Found an estimated cost of 25 for instruction: %ctlz
+; SSE42: Found an estimated cost of 23 for instruction: %ctlz
+; AVX: Found an estimated cost of 23 for instruction: %ctlz
+ %ctlz = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 1)
+ ret <2 x i64> %ctlz
+}
+
+define <4 x i64> @var_ctlz_v4i64(<4 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i64':
+; SSE2: Found an estimated cost of 50 for instruction: %ctlz
+; SSE42: Found an estimated cost of 46 for instruction: %ctlz
+; AVX1: Found an estimated cost of 48 for instruction: %ctlz
+; AVX2: Found an estimated cost of 23 for instruction: %ctlz
+ %ctlz = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> %a, i1 0)
+ ret <4 x i64> %ctlz
+}
+
+define <4 x i64> @var_ctlz_v4i64u(<4 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i64u':
+; SSE2: Found an estimated cost of 50 for instruction: %ctlz
+; SSE42: Found an estimated cost of 46 for instruction: %ctlz
+; AVX1: Found an estimated cost of 48 for instruction: %ctlz
+; AVX2: Found an estimated cost of 23 for instruction: %ctlz
+ %ctlz = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> %a, i1 1)
+ ret <4 x i64> %ctlz
+}
+
+define <4 x i32> @var_ctlz_v4i32(<4 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i32':
+; SSE2: Found an estimated cost of 26 for instruction: %ctlz
+; SSE42: Found an estimated cost of 18 for instruction: %ctlz
+; AVX: Found an estimated cost of 18 for instruction: %ctlz
+ %ctlz = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %a, i1 0)
+ ret <4 x i32> %ctlz
+}
+
+define <4 x i32> @var_ctlz_v4i32u(<4 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v4i32u':
+; SSE2: Found an estimated cost of 26 for instruction: %ctlz
+; SSE42: Found an estimated cost of 18 for instruction: %ctlz
+; AVX: Found an estimated cost of 18 for instruction: %ctlz
+ %ctlz = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %a, i1 1)
+ ret <4 x i32> %ctlz
+}
+
+define <8 x i32> @var_ctlz_v8i32(<8 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i32':
+; SSE2: Found an estimated cost of 52 for instruction: %ctlz
+; SSE42: Found an estimated cost of 36 for instruction: %ctlz
+; AVX1: Found an estimated cost of 38 for instruction: %ctlz
+; AVX2: Found an estimated cost of 18 for instruction: %ctlz
+ %ctlz = call <8 x i32> @llvm.ctlz.v8i32(<8 x i32> %a, i1 0)
+ ret <8 x i32> %ctlz
+}
+
+define <8 x i32> @var_ctlz_v8i32u(<8 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i32u':
+; SSE2: Found an estimated cost of 52 for instruction: %ctlz
+; SSE42: Found an estimated cost of 36 for instruction: %ctlz
+; AVX1: Found an estimated cost of 38 for instruction: %ctlz
+; AVX2: Found an estimated cost of 18 for instruction: %ctlz
+ %ctlz = call <8 x i32> @llvm.ctlz.v8i32(<8 x i32> %a, i1 1)
+ ret <8 x i32> %ctlz
+}
+
+define <8 x i16> @var_ctlz_v8i16(<8 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i16':
+; SSE2: Found an estimated cost of 20 for instruction: %ctlz
+; SSE42: Found an estimated cost of 14 for instruction: %ctlz
+; AVX: Found an estimated cost of 14 for instruction: %ctlz
+ %ctlz = call <8 x i16> @llvm.ctlz.v8i16(<8 x i16> %a, i1 0)
+ ret <8 x i16> %ctlz
+}
+
+define <8 x i16> @var_ctlz_v8i16u(<8 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v8i16u':
+; SSE2: Found an estimated cost of 20 for instruction: %ctlz
+; SSE42: Found an estimated cost of 14 for instruction: %ctlz
+; AVX: Found an estimated cost of 14 for instruction: %ctlz
+ %ctlz = call <8 x i16> @llvm.ctlz.v8i16(<8 x i16> %a, i1 1)
+ ret <8 x i16> %ctlz
+}
+
+define <16 x i16> @var_ctlz_v16i16(<16 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i16':
+; SSE2: Found an estimated cost of 40 for instruction: %ctlz
+; SSE42: Found an estimated cost of 28 for instruction: %ctlz
+; AVX1: Found an estimated cost of 30 for instruction: %ctlz
+; AVX2: Found an estimated cost of 14 for instruction: %ctlz
+ %ctlz = call <16 x i16> @llvm.ctlz.v16i16(<16 x i16> %a, i1 0)
+ ret <16 x i16> %ctlz
+}
+
+define <16 x i16> @var_ctlz_v16i16u(<16 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i16u':
+; SSE2: Found an estimated cost of 40 for instruction: %ctlz
+; SSE42: Found an estimated cost of 28 for instruction: %ctlz
+; AVX1: Found an estimated cost of 30 for instruction: %ctlz
+; AVX2: Found an estimated cost of 14 for instruction: %ctlz
+ %ctlz = call <16 x i16> @llvm.ctlz.v16i16(<16 x i16> %a, i1 1)
+ ret <16 x i16> %ctlz
+}
+
+define <16 x i8> @var_ctlz_v16i8(<16 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i8':
+; SSE2: Found an estimated cost of 17 for instruction: %ctlz
+; SSE42: Found an estimated cost of 9 for instruction: %ctlz
+; AVX: Found an estimated cost of 9 for instruction: %ctlz
+ %ctlz = call <16 x i8> @llvm.ctlz.v16i8(<16 x i8> %a, i1 0)
+ ret <16 x i8> %ctlz
+}
+
+define <16 x i8> @var_ctlz_v16i8u(<16 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v16i8u':
+; SSE2: Found an estimated cost of 17 for instruction: %ctlz
+; SSE42: Found an estimated cost of 9 for instruction: %ctlz
+; AVX: Found an estimated cost of 9 for instruction: %ctlz
+ %ctlz = call <16 x i8> @llvm.ctlz.v16i8(<16 x i8> %a, i1 1)
+ ret <16 x i8> %ctlz
+}
+
+define <32 x i8> @var_ctlz_v32i8(<32 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v32i8':
+; SSE2: Found an estimated cost of 34 for instruction: %ctlz
+; SSE42: Found an estimated cost of 18 for instruction: %ctlz
+; AVX1: Found an estimated cost of 20 for instruction: %ctlz
+; AVX2: Found an estimated cost of 9 for instruction: %ctlz
+ %ctlz = call <32 x i8> @llvm.ctlz.v32i8(<32 x i8> %a, i1 0)
+ ret <32 x i8> %ctlz
+}
+
+define <32 x i8> @var_ctlz_v32i8u(<32 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctlz_v32i8u':
+; SSE2: Found an estimated cost of 34 for instruction: %ctlz
+; SSE42: Found an estimated cost of 18 for instruction: %ctlz
+; AVX1: Found an estimated cost of 20 for instruction: %ctlz
+; AVX2: Found an estimated cost of 9 for instruction: %ctlz
+ %ctlz = call <32 x i8> @llvm.ctlz.v32i8(<32 x i8> %a, i1 1)
+ ret <32 x i8> %ctlz
+}
diff --git a/test/Analysis/CostModel/X86/ctpop.ll b/test/Analysis/CostModel/X86/ctpop.ll
new file mode 100644
index 0000000000000..f072cbaec4926
--- /dev/null
+++ b/test/Analysis/CostModel/X86/ctpop.ll
@@ -0,0 +1,133 @@
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=pentium4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE2 -check-prefix=NOPOPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE42 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+
+; Verify the cost of scalar population count instructions.
+
+declare i64 @llvm.ctpop.i64(i64)
+declare i32 @llvm.ctpop.i32(i32)
+declare i16 @llvm.ctpop.i16(i16)
+declare i8 @llvm.ctpop.i8(i8)
+
+define i64 @var_ctpop_i64(i64 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i64':
+; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
+; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
+ %ctpop = call i64 @llvm.ctpop.i64(i64 %a)
+ ret i64 %ctpop
+}
+
+define i32 @var_ctpop_i32(i32 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i32':
+; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
+; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
+ %ctpop = call i32 @llvm.ctpop.i32(i32 %a)
+ ret i32 %ctpop
+}
+
+define i16 @var_ctpop_i16(i16 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i16':
+; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
+; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
+ %ctpop = call i16 @llvm.ctpop.i16(i16 %a)
+ ret i16 %ctpop
+}
+
+define i8 @var_ctpop_i8(i8 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_i8':
+; NOPOPCNT: Found an estimated cost of 4 for instruction: %ctpop
+; POPCNT: Found an estimated cost of 1 for instruction: %ctpop
+ %ctpop = call i8 @llvm.ctpop.i8(i8 %a)
+ ret i8 %ctpop
+}
+
+; Verify the cost of vector population count instructions.
+
+declare <2 x i64> @llvm.ctpop.v2i64(<2 x i64>)
+declare <4 x i32> @llvm.ctpop.v4i32(<4 x i32>)
+declare <8 x i16> @llvm.ctpop.v8i16(<8 x i16>)
+declare <16 x i8> @llvm.ctpop.v16i8(<16 x i8>)
+
+declare <4 x i64> @llvm.ctpop.v4i64(<4 x i64>)
+declare <8 x i32> @llvm.ctpop.v8i32(<8 x i32>)
+declare <16 x i16> @llvm.ctpop.v16i16(<16 x i16>)
+declare <32 x i8> @llvm.ctpop.v32i8(<32 x i8>)
+
+define <2 x i64> @var_ctpop_v2i64(<2 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v2i64':
+; SSE2: Found an estimated cost of 12 for instruction: %ctpop
+; SSE42: Found an estimated cost of 7 for instruction: %ctpop
+; AVX: Found an estimated cost of 7 for instruction: %ctpop
+ %ctpop = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a)
+ ret <2 x i64> %ctpop
+}
+
+define <4 x i64> @var_ctpop_v4i64(<4 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v4i64':
+; SSE2: Found an estimated cost of 24 for instruction: %ctpop
+; SSE42: Found an estimated cost of 14 for instruction: %ctpop
+; AVX1: Found an estimated cost of 16 for instruction: %ctpop
+; AVX2: Found an estimated cost of 7 for instruction: %ctpop
+ %ctpop = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> %a)
+ ret <4 x i64> %ctpop
+}
+
+define <4 x i32> @var_ctpop_v4i32(<4 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v4i32':
+; SSE2: Found an estimated cost of 15 for instruction: %ctpop
+; SSE42: Found an estimated cost of 11 for instruction: %ctpop
+; AVX: Found an estimated cost of 11 for instruction: %ctpop
+ %ctpop = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> %a)
+ ret <4 x i32> %ctpop
+}
+
+define <8 x i32> @var_ctpop_v8i32(<8 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v8i32':
+; SSE2: Found an estimated cost of 30 for instruction: %ctpop
+; SSE42: Found an estimated cost of 22 for instruction: %ctpop
+; AVX1: Found an estimated cost of 24 for instruction: %ctpop
+; AVX2: Found an estimated cost of 11 for instruction: %ctpop
+ %ctpop = call <8 x i32> @llvm.ctpop.v8i32(<8 x i32> %a)
+ ret <8 x i32> %ctpop
+}
+
+define <8 x i16> @var_ctpop_v8i16(<8 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v8i16':
+; SSE2: Found an estimated cost of 13 for instruction: %ctpop
+; SSE42: Found an estimated cost of 9 for instruction: %ctpop
+; AVX: Found an estimated cost of 9 for instruction: %ctpop
+ %ctpop = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> %a)
+ ret <8 x i16> %ctpop
+}
+
+define <16 x i16> @var_ctpop_v16i16(<16 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v16i16':
+; SSE2: Found an estimated cost of 26 for instruction: %ctpop
+; SSE42: Found an estimated cost of 18 for instruction: %ctpop
+; AVX1: Found an estimated cost of 20 for instruction: %ctpop
+; AVX2: Found an estimated cost of 9 for instruction: %ctpop
+ %ctpop = call <16 x i16> @llvm.ctpop.v16i16(<16 x i16> %a)
+ ret <16 x i16> %ctpop
+}
+
+define <16 x i8> @var_ctpop_v16i8(<16 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v16i8':
+; SSE2: Found an estimated cost of 10 for instruction: %ctpop
+; SSE42: Found an estimated cost of 6 for instruction: %ctpop
+; AVX: Found an estimated cost of 6 for instruction: %ctpop
+ %ctpop = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> %a)
+ ret <16 x i8> %ctpop
+}
+
+define <32 x i8> @var_ctpop_v32i8(<32 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_ctpop_v32i8':
+; SSE2: Found an estimated cost of 20 for instruction: %ctpop
+; SSE42: Found an estimated cost of 12 for instruction: %ctpop
+; AVX1: Found an estimated cost of 14 for instruction: %ctpop
+; AVX2: Found an estimated cost of 6 for instruction: %ctpop
+ %ctpop = call <32 x i8> @llvm.ctpop.v32i8(<32 x i8> %a)
+ ret <32 x i8> %ctpop
+}
diff --git a/test/Analysis/CostModel/X86/cttz.ll b/test/Analysis/CostModel/X86/cttz.ll
new file mode 100644
index 0000000000000..5d3c59b602320
--- /dev/null
+++ b/test/Analysis/CostModel/X86/cttz.ll
@@ -0,0 +1,233 @@
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=pentium4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE2 -check-prefix=NOPOPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=SSE -check-prefix=SSE42 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7-avx -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=core-avx2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver2 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX1 -check-prefix=POPCNT
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=bdver4 -cost-model -analyze | FileCheck %s -check-prefix=CHECK -check-prefix=AVX -check-prefix=AVX2 -check-prefix=POPCNT
+
+; Verify the cost of scalar trailing zero count instructions.
+
+declare i64 @llvm.cttz.i64(i64, i1)
+declare i32 @llvm.cttz.i32(i32, i1)
+declare i16 @llvm.cttz.i16(i16, i1)
+declare i8 @llvm.cttz.i8(i8, i1)
+
+define i64 @var_cttz_i64(i64 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i64':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i64 @llvm.cttz.i64(i64 %a, i1 0)
+ ret i64 %cttz
+}
+
+define i64 @var_cttz_i64u(i64 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i64u':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i64 @llvm.cttz.i64(i64 %a, i1 1)
+ ret i64 %cttz
+}
+
+define i32 @var_cttz_i32(i32 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i32':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i32 @llvm.cttz.i32(i32 %a, i1 0)
+ ret i32 %cttz
+}
+
+define i32 @var_cttz_i32u(i32 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i32u':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i32 @llvm.cttz.i32(i32 %a, i1 1)
+ ret i32 %cttz
+}
+
+define i16 @var_cttz_i16(i16 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i16':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i16 @llvm.cttz.i16(i16 %a, i1 0)
+ ret i16 %cttz
+}
+
+define i16 @var_cttz_i16u(i16 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i16u':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i16 @llvm.cttz.i16(i16 %a, i1 1)
+ ret i16 %cttz
+}
+
+define i8 @var_cttz_i8(i8 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i8':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i8 @llvm.cttz.i8(i8 %a, i1 0)
+ ret i8 %cttz
+}
+
+define i8 @var_cttz_i8u(i8 %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_i8u':
+; CHECK: Found an estimated cost of 1 for instruction: %cttz
+ %cttz = call i8 @llvm.cttz.i8(i8 %a, i1 1)
+ ret i8 %cttz
+}
+
+; Verify the cost of vector trailing zero count instructions.
+
+declare <2 x i64> @llvm.cttz.v2i64(<2 x i64>, i1)
+declare <4 x i32> @llvm.cttz.v4i32(<4 x i32>, i1)
+declare <8 x i16> @llvm.cttz.v8i16(<8 x i16>, i1)
+declare <16 x i8> @llvm.cttz.v16i8(<16 x i8>, i1)
+
+declare <4 x i64> @llvm.cttz.v4i64(<4 x i64>, i1)
+declare <8 x i32> @llvm.cttz.v8i32(<8 x i32>, i1)
+declare <16 x i16> @llvm.cttz.v16i16(<16 x i16>, i1)
+declare <32 x i8> @llvm.cttz.v32i8(<32 x i8>, i1)
+
+define <2 x i64> @var_cttz_v2i64(<2 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v2i64':
+; SSE2: Found an estimated cost of 14 for instruction: %cttz
+; SSE42: Found an estimated cost of 10 for instruction: %cttz
+; AVX: Found an estimated cost of 10 for instruction: %cttz
+ %cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 0)
+ ret <2 x i64> %cttz
+}
+
+define <2 x i64> @var_cttz_v2i64u(<2 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v2i64u':
+; SSE2: Found an estimated cost of 14 for instruction: %cttz
+; SSE42: Found an estimated cost of 10 for instruction: %cttz
+; AVX: Found an estimated cost of 10 for instruction: %cttz
+ %cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 1)
+ ret <2 x i64> %cttz
+}
+
+define <4 x i64> @var_cttz_v4i64(<4 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i64':
+; SSE2: Found an estimated cost of 28 for instruction: %cttz
+; SSE42: Found an estimated cost of 20 for instruction: %cttz
+; AVX1: Found an estimated cost of 22 for instruction: %cttz
+; AVX2: Found an estimated cost of 10 for instruction: %cttz
+ %cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 0)
+ ret <4 x i64> %cttz
+}
+
+define <4 x i64> @var_cttz_v4i64u(<4 x i64> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i64u':
+; SSE2: Found an estimated cost of 28 for instruction: %cttz
+; SSE42: Found an estimated cost of 20 for instruction: %cttz
+; AVX1: Found an estimated cost of 22 for instruction: %cttz
+; AVX2: Found an estimated cost of 10 for instruction: %cttz
+ %cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 1)
+ ret <4 x i64> %cttz
+}
+
+define <4 x i32> @var_cttz_v4i32(<4 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i32':
+; SSE2: Found an estimated cost of 18 for instruction: %cttz
+; SSE42: Found an estimated cost of 14 for instruction: %cttz
+; AVX: Found an estimated cost of 14 for instruction: %cttz
+ %cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 0)
+ ret <4 x i32> %cttz
+}
+
+define <4 x i32> @var_cttz_v4i32u(<4 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v4i32u':
+; SSE2: Found an estimated cost of 18 for instruction: %cttz
+; SSE42: Found an estimated cost of 14 for instruction: %cttz
+; AVX: Found an estimated cost of 14 for instruction: %cttz
+ %cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 1)
+ ret <4 x i32> %cttz
+}
+
+define <8 x i32> @var_cttz_v8i32(<8 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i32':
+; SSE2: Found an estimated cost of 36 for instruction: %cttz
+; SSE42: Found an estimated cost of 28 for instruction: %cttz
+; AVX1: Found an estimated cost of 30 for instruction: %cttz
+; AVX2: Found an estimated cost of 14 for instruction: %cttz
+ %cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 0)
+ ret <8 x i32> %cttz
+}
+
+define <8 x i32> @var_cttz_v8i32u(<8 x i32> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i32u':
+; SSE2: Found an estimated cost of 36 for instruction: %cttz
+; SSE42: Found an estimated cost of 28 for instruction: %cttz
+; AVX1: Found an estimated cost of 30 for instruction: %cttz
+; AVX2: Found an estimated cost of 14 for instruction: %cttz
+ %cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 1)
+ ret <8 x i32> %cttz
+}
+
+define <8 x i16> @var_cttz_v8i16(<8 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i16':
+; SSE2: Found an estimated cost of 16 for instruction: %cttz
+; SSE42: Found an estimated cost of 12 for instruction: %cttz
+; AVX: Found an estimated cost of 12 for instruction: %cttz
+ %cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 0)
+ ret <8 x i16> %cttz
+}
+
+define <8 x i16> @var_cttz_v8i16u(<8 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v8i16u':
+; SSE2: Found an estimated cost of 16 for instruction: %cttz
+; SSE42: Found an estimated cost of 12 for instruction: %cttz
+; AVX: Found an estimated cost of 12 for instruction: %cttz
+ %cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 1)
+ ret <8 x i16> %cttz
+}
+
+define <16 x i16> @var_cttz_v16i16(<16 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i16':
+; SSE2: Found an estimated cost of 32 for instruction: %cttz
+; SSE42: Found an estimated cost of 24 for instruction: %cttz
+; AVX1: Found an estimated cost of 26 for instruction: %cttz
+; AVX2: Found an estimated cost of 12 for instruction: %cttz
+ %cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 0)
+ ret <16 x i16> %cttz
+}
+
+define <16 x i16> @var_cttz_v16i16u(<16 x i16> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i16u':
+; SSE2: Found an estimated cost of 32 for instruction: %cttz
+; SSE42: Found an estimated cost of 24 for instruction: %cttz
+; AVX1: Found an estimated cost of 26 for instruction: %cttz
+; AVX2: Found an estimated cost of 12 for instruction: %cttz
+ %cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 1)
+ ret <16 x i16> %cttz
+}
+
+define <16 x i8> @var_cttz_v16i8(<16 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i8':
+; SSE2: Found an estimated cost of 13 for instruction: %cttz
+; SSE42: Found an estimated cost of 9 for instruction: %cttz
+; AVX: Found an estimated cost of 9 for instruction: %cttz
+ %cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 0)
+ ret <16 x i8> %cttz
+}
+
+define <16 x i8> @var_cttz_v16i8u(<16 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v16i8u':
+; SSE2: Found an estimated cost of 13 for instruction: %cttz
+; SSE42: Found an estimated cost of 9 for instruction: %cttz
+; AVX: Found an estimated cost of 9 for instruction: %cttz
+ %cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 1)
+ ret <16 x i8> %cttz
+}
+
+define <32 x i8> @var_cttz_v32i8(<32 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v32i8':
+; SSE2: Found an estimated cost of 26 for instruction: %cttz
+; SSE42: Found an estimated cost of 18 for instruction: %cttz
+; AVX1: Found an estimated cost of 20 for instruction: %cttz
+; AVX2: Found an estimated cost of 9 for instruction: %cttz
+ %cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 0)
+ ret <32 x i8> %cttz
+}
+
+define <32 x i8> @var_cttz_v32i8u(<32 x i8> %a) {
+; CHECK: 'Cost Model Analysis' for function 'var_cttz_v32i8u':
+; SSE2: Found an estimated cost of 26 for instruction: %cttz
+; SSE42: Found an estimated cost of 18 for instruction: %cttz
+; AVX1: Found an estimated cost of 20 for instruction: %cttz
+; AVX2: Found an estimated cost of 9 for instruction: %cttz
+ %cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 1)
+ ret <32 x i8> %cttz
+}
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-03 03:09:45 +0000