1 ; RUN: opt < %s -cost-model -analyze | FileCheck %s -D#VBITS=128
2 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=128 | FileCheck %s -D#VBITS=128
3 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=256 | FileCheck %s -D#VBITS=256
4 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=384 | FileCheck %s -D#VBITS=256
5 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=512 | FileCheck %s -D#VBITS=512
6 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=640 | FileCheck %s -D#VBITS=512
7 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=768 | FileCheck %s -D#VBITS=512
8 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=896 | FileCheck %s -D#VBITS=512
9 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1024 | FileCheck %s -D#VBITS=1024
10 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1152 | FileCheck %s -D#VBITS=1024
11 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1280 | FileCheck %s -D#VBITS=1024
12 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1408 | FileCheck %s -D#VBITS=1024
13 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1536 | FileCheck %s -D#VBITS=1024
14 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1664 | FileCheck %s -D#VBITS=1024
15 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1792 | FileCheck %s -D#VBITS=1024
16 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1920 | FileCheck %s -D#VBITS=1024
17 ; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=2048 | FileCheck %s -D#VBITS=2048
19 ; VBITS represents the useful bit size of a vector register from the code
20 ; generator's point of view. It is clamped to power-of-2 values because
21 ; only power-of-2 vector lengths are considered legal, regardless of the
22 ; user specified vector length.
24 target triple = "aarch64-unknown-linux-gnu"
26 ; Ensure the cost of legalisation is removed as the vector length grows.
27 ; NOTE: Assumes BaseCost_add=1, BaseCost_fadd=2.
28 define void @add() #0 {
29 ; CHECK-LABEL: Printing analysis 'Cost Model Analysis' for function 'add':
30 ; CHECK: cost of [[#div(127,VBITS)+1]] for instruction: %add128 = add <4 x i32> undef, undef
31 ; CHECK: cost of [[#div(255,VBITS)+1]] for instruction: %add256 = add <8 x i32> undef, undef
32 ; CHECK: cost of [[#div(511,VBITS)+1]] for instruction: %add512 = add <16 x i32> undef, undef
33 ; CHECK: cost of [[#div(1023,VBITS)+1]] for instruction: %add1024 = add <32 x i32> undef, undef
34 ; CHECK: cost of [[#div(2047,VBITS)+1]] for instruction: %add2048 = add <64 x i32> undef, undef
35 %add128 = add <4 x i32> undef, undef
36 %add256 = add <8 x i32> undef, undef
37 %add512 = add <16 x i32> undef, undef
38 %add1024 = add <32 x i32> undef, undef
39 %add2048 = add <64 x i32> undef, undef
41 ; Using a single vector length, ensure all element types are recognised.
42 ; CHECK: cost of [[#div(511,VBITS)+1]] for instruction: %add512.i8 = add <64 x i8> undef, undef
43 ; CHECK: cost of [[#div(511,VBITS)+1]] for instruction: %add512.i16 = add <32 x i16> undef, undef
44 ; CHECK: cost of [[#div(511,VBITS)+1]] for instruction: %add512.i32 = add <16 x i32> undef, undef
45 ; CHECK: cost of [[#div(511,VBITS)+1]] for instruction: %add512.i64 = add <8 x i64> undef, undef
46 ; CHECK: cost of [[#mul(div(511,VBITS)+1,2)]] for instruction: %add512.f16 = fadd <32 x half> undef, undef
47 ; CHECK: cost of [[#mul(div(511,VBITS)+1,2)]] for instruction: %add512.f32 = fadd <16 x float> undef, undef
48 ; CHECK: cost of [[#mul(div(511,VBITS)+1,2)]] for instruction: %add512.f64 = fadd <8 x double> undef, undef
49 %add512.i8 = add <64 x i8> undef, undef
50 %add512.i16 = add <32 x i16> undef, undef
51 %add512.i32 = add <16 x i32> undef, undef
52 %add512.i64 = add <8 x i64> undef, undef
53 %add512.f16 = fadd <32 x half> undef, undef
54 %add512.f32 = fadd <16 x float> undef, undef
55 %add512.f64 = fadd <8 x double> undef, undef
60 attributes #0 = { "target-features"="+sve" }