maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AMDGPU/optimize-negated-cond.ll

76 lines
1.7 KiB
LLVM
Raw Normal View History

[AMDGPU] Simplify negated condition Optimize sequence: %sel = V_CNDMASK_B32_e64 0, 1, %cc %cmp = V_CMP_NE_U32 1, %1 $vcc = S_AND_B64 $exec, %cmp S_CBRANCH_VCC[N]Z => $vcc = S_ANDN2_B64 $exec, %cc S_CBRANCH_VCC[N]Z It is the negation pattern inserted by DAGCombiner::visitBRCOND() in the rebuildSetCC(). Differential Revision: https://reviews.llvm.org/D55402 llvm-svn: 349003
2018-12-13 11:17:40 +08:00
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
; GCN-LABEL: {{^}}negated_cond:
; GCN: BB0_1:
; GCN: v_cmp_eq_u32_e64 [[CC:[^,]+]],
[MBP] Move a latch block with conditional exit and multi predecessors to top of loop Current findBestLoopTop can find and move one kind of block to top, a latch block has one successor. Another common case is: * a latch block * it has two successors, one is loop header, another is exit * it has more than one predecessors If it is below one of its predecessors P, only P can fall through to it, all other predecessors need a jump to it, and another conditional jump to loop header. If it is moved before loop header, all its predecessors jump to it, then fall through to loop header. So all its predecessors except P can reduce one taken branch. Differential Revision: https://reviews.llvm.org/D43256 llvm-svn: 363471
2019-06-15 07:08:59 +08:00
; GCN: BB0_3:
[AMDGPU] Simplify negated condition Optimize sequence: %sel = V_CNDMASK_B32_e64 0, 1, %cc %cmp = V_CMP_NE_U32 1, %1 $vcc = S_AND_B64 $exec, %cmp S_CBRANCH_VCC[N]Z => $vcc = S_ANDN2_B64 $exec, %cc S_CBRANCH_VCC[N]Z It is the negation pattern inserted by DAGCombiner::visitBRCOND() in the rebuildSetCC(). Differential Revision: https://reviews.llvm.org/D55402 llvm-svn: 349003
2018-12-13 11:17:40 +08:00
; GCN-NOT: v_cndmask_b32
; GCN-NOT: v_cmp
; GCN: s_andn2_b64 vcc, exec, [[CC]]
[MBP] Move a latch block with conditional exit and multi predecessors to top of loop Current findBestLoopTop can find and move one kind of block to top, a latch block has one successor. Another common case is: * a latch block * it has two successors, one is loop header, another is exit * it has more than one predecessors If it is below one of its predecessors P, only P can fall through to it, all other predecessors need a jump to it, and another conditional jump to loop header. If it is moved before loop header, all its predecessors jump to it, then fall through to loop header. So all its predecessors except P can reduce one taken branch. Differential Revision: https://reviews.llvm.org/D43256 llvm-svn: 363471
2019-06-15 07:08:59 +08:00
; GCN: s_cbranch_vccnz BB0_2
[AMDGPU] Simplify negated condition Optimize sequence: %sel = V_CNDMASK_B32_e64 0, 1, %cc %cmp = V_CMP_NE_U32 1, %1 $vcc = S_AND_B64 $exec, %cmp S_CBRANCH_VCC[N]Z => $vcc = S_ANDN2_B64 $exec, %cc S_CBRANCH_VCC[N]Z It is the negation pattern inserted by DAGCombiner::visitBRCOND() in the rebuildSetCC(). Differential Revision: https://reviews.llvm.org/D55402 llvm-svn: 349003
2018-12-13 11:17:40 +08:00
define amdgpu_kernel void @negated_cond(i32 addrspace(1)* %arg1) {
bb:
br label %bb1
bb1:
%tmp1 = load i32, i32 addrspace(1)* %arg1
%tmp2 = icmp eq i32 %tmp1, 0
br label %bb2
bb2:
%tmp3 = phi i32 [ 0, %bb1 ], [ %tmp6, %bb4 ]
%tmp4 = shl i32 %tmp3, 5
br i1 %tmp2, label %bb3, label %bb4
bb3:
%tmp5 = add i32 %tmp4, 1
br label %bb4
bb4:
%tmp6 = phi i32 [ %tmp5, %bb3 ], [ %tmp4, %bb2 ]
%gep = getelementptr inbounds i32, i32 addrspace(1)* %arg1, i32 %tmp6
store i32 0, i32 addrspace(1)* %gep
%tmp7 = icmp eq i32 %tmp6, 32
br i1 %tmp7, label %bb1, label %bb2
}
; GCN-LABEL: {{^}}negated_cond_dominated_blocks:
; GCN: v_cmp_eq_u32_e64 [[CC:[^,]+]],
[MBP] Move a latch block with conditional exit and multi predecessors to top of loop Current findBestLoopTop can find and move one kind of block to top, a latch block has one successor. Another common case is: * a latch block * it has two successors, one is loop header, another is exit * it has more than one predecessors If it is below one of its predecessors P, only P can fall through to it, all other predecessors need a jump to it, and another conditional jump to loop header. If it is moved before loop header, all its predecessors jump to it, then fall through to loop header. So all its predecessors except P can reduce one taken branch. Differential Revision: https://reviews.llvm.org/D43256 llvm-svn: 363471
2019-06-15 07:08:59 +08:00
; GCN: %bb4
[AMDGPU] Simplify negated condition Optimize sequence: %sel = V_CNDMASK_B32_e64 0, 1, %cc %cmp = V_CMP_NE_U32 1, %1 $vcc = S_AND_B64 $exec, %cmp S_CBRANCH_VCC[N]Z => $vcc = S_ANDN2_B64 $exec, %cc S_CBRANCH_VCC[N]Z It is the negation pattern inserted by DAGCombiner::visitBRCOND() in the rebuildSetCC(). Differential Revision: https://reviews.llvm.org/D55402 llvm-svn: 349003
2018-12-13 11:17:40 +08:00
; GCN-NOT: v_cndmask_b32
; GCN-NOT: v_cmp
; GCN: s_andn2_b64 vcc, exec, [[CC]]
[MBP] Move a latch block with conditional exit and multi predecessors to top of loop Current findBestLoopTop can find and move one kind of block to top, a latch block has one successor. Another common case is: * a latch block * it has two successors, one is loop header, another is exit * it has more than one predecessors If it is below one of its predecessors P, only P can fall through to it, all other predecessors need a jump to it, and another conditional jump to loop header. If it is moved before loop header, all its predecessors jump to it, then fall through to loop header. So all its predecessors except P can reduce one taken branch. Differential Revision: https://reviews.llvm.org/D43256 llvm-svn: 363471
2019-06-15 07:08:59 +08:00
; GCN: s_cbranch_vccnz BB1_1
[AMDGPU] Simplify negated condition Optimize sequence: %sel = V_CNDMASK_B32_e64 0, 1, %cc %cmp = V_CMP_NE_U32 1, %1 $vcc = S_AND_B64 $exec, %cmp S_CBRANCH_VCC[N]Z => $vcc = S_ANDN2_B64 $exec, %cc S_CBRANCH_VCC[N]Z It is the negation pattern inserted by DAGCombiner::visitBRCOND() in the rebuildSetCC(). Differential Revision: https://reviews.llvm.org/D55402 llvm-svn: 349003
2018-12-13 11:17:40 +08:00
define amdgpu_kernel void @negated_cond_dominated_blocks(i32 addrspace(1)* %arg1) {
bb:
br label %bb2
bb2:
%tmp1 = load i32, i32 addrspace(1)* %arg1
%tmp2 = icmp eq i32 %tmp1, 0
br label %bb4
bb3:
ret void
bb4:
%tmp3 = phi i32 [ 0, %bb2 ], [ %tmp7, %bb7 ]
%tmp4 = shl i32 %tmp3, 5
br i1 %tmp2, label %bb5, label %bb6
bb5:
%tmp5 = add i32 %tmp4, 1
br label %bb7
bb6:
%tmp6 = add i32 %tmp3, 1
br label %bb7
bb7:
%tmp7 = phi i32 [ %tmp5, %bb5 ], [ %tmp6, %bb6 ]
%gep = getelementptr inbounds i32, i32 addrspace(1)* %arg1, i32 %tmp7
store i32 0, i32 addrspace(1)* %gep
%tmp8 = icmp eq i32 %tmp7, 32
br i1 %tmp8, label %bb3, label %bb4
}