maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[ARM][MVE] Select vqneg 

Adds a pattern to ARMInstrMVE.td to use a VQNEG
  instruction if an equivalent multi-instruction
  construct is found.

Differential Revision: https://reviews.llvm.org/D70491
This commit is contained in:
Anna Welker 2019-11-25 11:17:13 +00:00
parent 2de23c8364
commit 6fc3e6f2eb
2 changed files with 63 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
llvm/lib/Target/ARM/ARMInstrMVE.td
View File

@ -1961,12 +1961,13 @@ def MVE_VQNEGs32 : MVE_VQABSNEG<"vqneg", "s32", 0b10, 0b1>;
// int_min/int_max: vector containing INT_MIN/INT_MAX VTI.Size times
// zero_vec: v4i32-initialized zero vector, potentially wrapped in a bitconvert
multiclass vqabs_pattern<MVEVectorVTInfo VTI, dag int_min, dag int_max,
dag zero_vec, MVE_VQABSNEG vqabs_instruction> {
// The below tree can be replaced by a vqabs instruction, as it represents
// the following vectorized expression (r being the value in $reg):
// r > 0 ? r : (r == INT_MIN ? INT_MAX : -r)
multiclass vqabsneg_pattern<MVEVectorVTInfo VTI, dag int_min, dag int_max,
dag zero_vec, MVE_VQABSNEG vqabs_instruction,
MVE_VQABSNEG vqneg_instruction> {
let Predicates = [HasMVEInt] in {
// The below tree can be replaced by a vqabs instruction, as it represents
// the following vectorized expression (r being the value in $reg):
// r > 0 ? r : (r == INT_MIN ? INT_MAX : -r)
def : Pat<(VTI.Vec (vselect
(VTI.Pred (ARMvcmpz (VTI.Vec MQPR:$reg), (i32 12))),
(VTI.Vec MQPR:$reg),
@ -1975,24 +1976,31 @@ multiclass vqabs_pattern<MVEVectorVTInfo VTI, dag int_min, dag int_max,
int_max,
(sub (VTI.Vec zero_vec), (VTI.Vec MQPR:$reg)))))),
(VTI.Vec (vqabs_instruction (VTI.Vec MQPR:$reg)))>;
// Similarly, this tree represents vqneg, i.e. the following vectorized expression:
// r == INT_MIN ? INT_MAX : -r
def : Pat<(VTI.Vec (vselect
(VTI.Pred (ARMvcmp (VTI.Vec MQPR:$reg), int_min, (i32 0))),
int_max,
(sub (VTI.Vec zero_vec), (VTI.Vec MQPR:$reg)))),
(VTI.Vec (vqneg_instruction (VTI.Vec MQPR:$reg)))>;
}
}
defm MVE_VQABS_Ps8 : vqabs_pattern<MVE_v16i8,
defm MVE_VQABSNEG_Ps8 : vqabsneg_pattern<MVE_v16i8,
(v16i8 (ARMvmovImm (i32 3712))),
(v16i8 (ARMvmovImm (i32 3711))),
(bitconvert (v4i32 (ARMvmovImm (i32 0)))),
MVE_VQABSs8>;
defm MVE_VQABS_Ps16 : vqabs_pattern<MVE_v8i16,
MVE_VQABSs8, MVE_VQNEGs8>;
defm MVE_VQABSNEG_Ps16 : vqabsneg_pattern<MVE_v8i16,
(v8i16 (ARMvmovImm (i32 2688))),
(v8i16 (ARMvmvnImm (i32 2688))),
(bitconvert (v4i32 (ARMvmovImm (i32 0)))),
MVE_VQABSs16>;
defm MVE_VQABS_Ps32 : vqabs_pattern<MVE_v4i32,
MVE_VQABSs16, MVE_VQNEGs16>;
defm MVE_VQABSNEG_Ps32 : vqabsneg_pattern<MVE_v4i32,
(v4i32 (ARMvmovImm (i32 1664))),
(v4i32 (ARMvmvnImm (i32 1664))),
(ARMvmovImm (i32 0)),
MVE_VQABSs32>;
MVE_VQABSs32, MVE_VQNEGs32>;
class MVE_mod_imm<string iname, string suffix, bits<4> cmode, bit op,
dag iops, list<dag> pattern=[]>

44
llvm/test/CodeGen/Thumb2/vqneg.ll Normal file
View File

@ -0,0 +1,44 @@
; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve %s -o - | FileCheck %s
define arm_aapcs_vfpcc <16 x i8> @vqneg_test16(<16 x i8> %A) nounwind {
; CHECK-LABEL: vqneg_test16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vqneg.s8 q0, q0
; CHECK-NEXT: bx lr
entry:
%0 = icmp eq <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>
%1 = sub nsw <16 x i8> zeroinitializer, %A
%2 = select <16 x i1> %0, <16 x i8> <i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127>, <16 x i8> %1
ret <16 x i8> %2
}
define arm_aapcs_vfpcc <8 x i16> @vqneg_test8(<8 x i16> %A) nounwind {
; CHECK-LABEL: vqneg_test8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vqneg.s16 q0, q0
; CHECK-NEXT: bx lr
entry:
%0 = icmp eq <8 x i16> %A, <i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768>
%1 = sub nsw <8 x i16> zeroinitializer, %A
%2 = select <8 x i1> %0, <8 x i16> <i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767>, <8 x i16> %1
ret <8 x i16> %2
}
define arm_aapcs_vfpcc <4 x i32> @vqneg_test4(<4 x i32> %A) nounwind {
; CHECK-LABEL: vqneg_test4:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vqneg.s32 q0, q0
; CHECK-NEXT: bx lr
entry:
%0 = icmp eq <4 x i32> %A, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
%1 = sub nsw <4 x i32> zeroinitializer, %A
%2 = select <4 x i1> %0, <4 x i32> <i32 2147483647, i32 2147483647, i32 2147483647, i32 2147483647>, <4 x i32> %1
ret <4 x i32> %2
}