RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-25 05:52:53 +02:00
Code Issues Projects Releases Wiki Activity

[CostModel][X86] Drop separate AVX512VL checks - they match existing AVX512 costs

Browse Source 
Keep the tests though.

llvm-svn: 292076
This commit is contained in:
Simon Pilgrim 2017-01-15 20:19:28 +00:00
parent 11e498d0eb
commit 1634b52e50
3 changed files with 9 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
test/Analysis/CostModel/X86/vshift-ashr-cost.ll
View File

@ -7,9 +7,9 @@
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BWVL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; Verify the cost of vector arithmetic shift right instructions. ; Verify the cost of vector arithmetic shift right instructions.
@ -124,8 +124,6 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift
; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 20 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = ashr <32 x i16> %a, %b %shift = ashr <32 x i16> %a, %b
ret <32 x i16> %shift ret <32 x i16> %shift
@ -151,8 +149,6 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; AVX2: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 24 for instruction: %shift
; AVX512F: Found an estimated cost of 24 for instruction: %shift ; AVX512F: Found an estimated cost of 24 for instruction: %shift
; AVX512BW: Found an estimated cost of 24 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift
; AVX512VL: Found an estimated cost of 24 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 24 for instruction: %shift
; XOP: Found an estimated cost of 4 for instruction: %shift ; XOP: Found an estimated cost of 4 for instruction: %shift
%shift = ashr <32 x i8> %a, %b %shift = ashr <32 x i8> %a, %b
ret <32 x i8> %shift ret <32 x i8> %shift
@ -166,8 +162,6 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift
; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift
; AVX512BW: Found an estimated cost of 24 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift
; AVX512VL: Found an estimated cost of 48 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 24 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = ashr <64 x i8> %a, %b %shift = ashr <64 x i8> %a, %b
ret <64 x i8> %shift ret <64 x i8> %shift
@ -346,8 +340,6 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift
; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift
; AVX512BW: Found an estimated cost of 24 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift
; AVX512VL: Found an estimated cost of 48 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 24 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%insert = insertelement <64 x i8> undef, i8 %b, i32 0 %insert = insertelement <64 x i8> undef, i8 %b, i32 0
%splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
@ -466,8 +458,6 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift
; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 20 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> %shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
ret <32 x i16> %shift ret <32 x i16> %shift
@ -505,8 +495,6 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift
; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift
; AVX512BW: Found an estimated cost of 24 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift
; AVX512VL: Found an estimated cost of 48 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 24 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> %shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
ret <64 x i8> %shift ret <64 x i8> %shift
@ -623,8 +611,6 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift
; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 2 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX: Found an estimated cost of 8 for instruction: %shift
; XOPAVX2: Found an estimated cost of 2 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift
%shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
@ -664,8 +650,6 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 8 for instruction: %shift
; AVX512F: Found an estimated cost of 8 for instruction: %shift ; AVX512F: Found an estimated cost of 8 for instruction: %shift
; AVX512BW: Found an estimated cost of 4 for instruction: %shift ; AVX512BW: Found an estimated cost of 4 for instruction: %shift
; AVX512VL: Found an estimated cost of 8 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 4 for instruction: %shift
; XOPAVX: Found an estimated cost of 16 for instruction: %shift ; XOPAVX: Found an estimated cost of 16 for instruction: %shift
; XOPAVX2: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 8 for instruction: %shift
%shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>

20
test/Analysis/CostModel/X86/vshift-lshr-cost.ll
View File

@ -7,9 +7,9 @@
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BWVL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; Verify the cost of vector logical shift right instructions. ; Verify the cost of vector logical shift right instructions.
@ -127,8 +127,6 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift
; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 20 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = lshr <32 x i16> %a, %b %shift = lshr <32 x i16> %a, %b
ret <32 x i16> %shift ret <32 x i16> %shift
@ -166,8 +164,6 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = lshr <64 x i8> %a, %b %shift = lshr <64 x i8> %a, %b
ret <64 x i8> %shift ret <64 x i8> %shift
@ -346,8 +342,6 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%insert = insertelement <64 x i8> undef, i8 %b, i32 0 %insert = insertelement <64 x i8> undef, i8 %b, i32 0
%splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
@ -469,8 +463,6 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift
; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 20 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> %shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
ret <32 x i16> %shift ret <32 x i16> %shift
@ -508,8 +500,6 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 8 for instruction: %shift ; XOP: Found an estimated cost of 8 for instruction: %shift
%shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> %shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
ret <64 x i8> %shift ret <64 x i8> %shift
@ -628,8 +618,6 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift
; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 2 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX: Found an estimated cost of 8 for instruction: %shift
; XOPAVX2: Found an estimated cost of 2 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift
%shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> %shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
@ -669,8 +657,6 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift
; AVX512F: Found an estimated cost of 4 for instruction: %shift ; AVX512F: Found an estimated cost of 4 for instruction: %shift
; AVX512BW: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 2 for instruction: %shift
; AVX512VL: Found an estimated cost of 4 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 2 for instruction: %shift
; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX: Found an estimated cost of 8 for instruction: %shift
; XOPAVX2: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 4 for instruction: %shift
%shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> %shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>

18
test/Analysis/CostModel/X86/vshift-shl-cost.ll
View File

@ -7,9 +7,9 @@
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512VL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512dq,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BWVL ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512bw,+avx512vl -cost-model -analyze | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512 --check-prefix=AVX512BW
; Verify the cost of vector shift left instructions. ; Verify the cost of vector shift left instructions.
@ -165,8 +165,6 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 4 for instruction: %shift ; XOP: Found an estimated cost of 4 for instruction: %shift
%shift = shl <64 x i8> %a, %b %shift = shl <64 x i8> %a, %b
ret <64 x i8> %shift ret <64 x i8> %shift
@ -345,8 +343,6 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 4 for instruction: %shift ; XOP: Found an estimated cost of 4 for instruction: %shift
%insert = insertelement <64 x i8> undef, i8 %b, i32 0 %insert = insertelement <64 x i8> undef, i8 %b, i32 0
%splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
@ -469,8 +465,6 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift
; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 2 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX: Found an estimated cost of 4 for instruction: %shift
; XOPAVX2: Found an estimated cost of 2 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift
%shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> %shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
@ -509,8 +503,6 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift
; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift
; AVX512BW: Found an estimated cost of 11 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift
; AVX512VL: Found an estimated cost of 22 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 11 for instruction: %shift
; XOP: Found an estimated cost of 4 for instruction: %shift ; XOP: Found an estimated cost of 4 for instruction: %shift
%shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> %shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
ret <64 x i8> %shift ret <64 x i8> %shift
@ -631,8 +623,6 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift
; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift
; AVX512BW: Found an estimated cost of 1 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift
; AVX512VL: Found an estimated cost of 2 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 1 for instruction: %shift
; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX: Found an estimated cost of 4 for instruction: %shift
; XOPAVX2: Found an estimated cost of 2 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift
%shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> %shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
@ -672,8 +662,6 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift
; AVX512F: Found an estimated cost of 4 for instruction: %shift ; AVX512F: Found an estimated cost of 4 for instruction: %shift
; AVX512BW: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 2 for instruction: %shift
; AVX512VL: Found an estimated cost of 4 for instruction: %shift
; AVX512BWVL: Found an estimated cost of 2 for instruction: %shift
; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX: Found an estimated cost of 8 for instruction: %shift
; XOPAVX2: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 4 for instruction: %shift
%shift = shl <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> %shift = shl <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>