RPCS3/llvm
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm.git synced 2026-07-01 21:04:04 -04:00
Code Issues Packages Projects Releases 10 Wiki Activity

[X86] Emit KTEST when possible

Browse Source 
https://reviews.llvm.org/D69111

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375197 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
David Zarzycki
2019-10-18 03:45:52 +00:00
parent 6960eb5d78
commit 3a2363c797
2 changed files with 29 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/Target/X86/X86ISelLowering.cpp
+23 -8
View File
@@ -20923,11 +20923,11 @@ static SDValue LowerVSETCC(SDValue Op, const X86Subtarget &Subtarget,
return Result;
}
// Try to select this as a KORTEST+SETCC if possible.
static SDValue EmitKORTEST(SDValue Op0, SDValue Op1, ISD::CondCode CC,
const SDLoc &dl, SelectionDAG &DAG,
const X86Subtarget &Subtarget,
SDValue &X86CC) {
// Try to select this as a KORTEST+SETCC or KTEST+SETCC if possible.
static SDValue EmitAVX512Test(SDValue Op0, SDValue Op1, ISD::CondCode CC,
const SDLoc &dl, SelectionDAG &DAG,
const X86Subtarget &Subtarget,
SDValue &X86CC) {
// Only support equality comparisons.
if (CC != ISD::SETEQ && CC != ISD::SETNE)
return SDValue();
@@ -20952,6 +20952,21 @@ static SDValue EmitKORTEST(SDValue Op0, SDValue Op1, ISD::CondCode CC,
} else
return SDValue();
// If the input is an AND, we can combine it's operands into the KTEST.
bool KTestable = false;
if (Subtarget.hasDQI() && (VT == MVT::v8i1 || VT == MVT::v16i1))
KTestable = true;
if (Subtarget.hasBWI() && (VT == MVT::v32i1 || VT == MVT::v64i1))
KTestable = true;
if (!isNullConstant(Op1))
KTestable = false;
if (KTestable && Op0.getOpcode() == ISD::AND && Op0.hasOneUse()) {
SDValue LHS = Op0.getOperand(0);
SDValue RHS = Op0.getOperand(1);
X86CC = DAG.getTargetConstant(X86Cond, dl, MVT::i8);
return DAG.getNode(X86ISD::KTEST, dl, MVT::i32, LHS, RHS);
}
// If the input is an OR, we can combine it's operands into the KORTEST.
SDValue LHS = Op0;
SDValue RHS = Op0;
@@ -20988,9 +21003,9 @@ SDValue X86TargetLowering::emitFlagsForSetcc(SDValue Op0, SDValue Op1,
return PTEST;
}
// Try to lower using KORTEST.
if (SDValue KORTEST = EmitKORTEST(Op0, Op1, CC, dl, DAG, Subtarget, X86CC))
return KORTEST;
// Try to lower using KORTEST or KTEST.
if (SDValue Test = EmitAVX512Test(Op0, Op1, CC, dl, DAG, Subtarget, X86CC))
return Test;
// Look for X == 0, X == 1, X != 0, or X != 1. We can simplify some forms of
// these.
test/CodeGen/X86/avx512-mask-op.ll
+6 -6
View File
@@ -2027,8 +2027,8 @@ define void @ktest_1(<8 x double> %in, double * %base) {
; SKX: ## %bb.0:
; SKX-NEXT: vcmpgtpd (%rdi), %zmm0, %k1
; SKX-NEXT: vmovupd 8(%rdi), %zmm1 {%k1} {z}
; SKX-NEXT: vcmpltpd %zmm1, %zmm0, %k0 {%k1}
; SKX-NEXT: kortestb %k0, %k0
; SKX-NEXT: vcmpltpd %zmm1, %zmm0, %k0
; SKX-NEXT: ktestb %k0, %k1
; SKX-NEXT: je LBB43_2
; SKX-NEXT: ## %bb.1: ## %L1
; SKX-NEXT: vmovapd %zmm0, (%rdi)
@@ -2060,8 +2060,8 @@ define void @ktest_1(<8 x double> %in, double * %base) {
; AVX512DQ: ## %bb.0:
; AVX512DQ-NEXT: vcmpgtpd (%rdi), %zmm0, %k1
; AVX512DQ-NEXT: vmovupd 8(%rdi), %zmm1 {%k1} {z}
; AVX512DQ-NEXT: vcmpltpd %zmm1, %zmm0, %k0 {%k1}
; AVX512DQ-NEXT: kortestb %k0, %k0
; AVX512DQ-NEXT: vcmpltpd %zmm1, %zmm0, %k0
; AVX512DQ-NEXT: ktestb %k0, %k1
; AVX512DQ-NEXT: je LBB43_2
; AVX512DQ-NEXT: ## %bb.1: ## %L1
; AVX512DQ-NEXT: vmovapd %zmm0, (%rdi)
@@ -2077,8 +2077,8 @@ define void @ktest_1(<8 x double> %in, double * %base) {
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: vcmpgtpd (%eax), %zmm0, %k1
; X86-NEXT: vmovupd 8(%eax), %zmm1 {%k1} {z}
; X86-NEXT: vcmpltpd %zmm1, %zmm0, %k0 {%k1}
; X86-NEXT: kortestb %k0, %k0
; X86-NEXT: vcmpltpd %zmm1, %zmm0, %k0
; X86-NEXT: ktestb %k0, %k1
; X86-NEXT: je LBB43_2
; X86-NEXT: ## %bb.1: ## %L1
; X86-NEXT: vmovapd %zmm0, (%eax)