[llvm] r351139 - [AArch64] Add new target feature to fuse arithmetic and logic operations
Evandro Menezes via llvm-commits
llvm-commits at lists.llvm.org
Mon Jan 14 17:04:03 PST 2019
Just committed rL351144 fixing this.
Thanks for the heads up, Reid.
--
Evandro Menezes
On 01/14/19 18:54, Reid Kleckner wrote:
> The lit internal shell does not support '|&':
>
> http://lab.llvm.org:8011/builders/clang-x64-windows-msvc/builds/3329
> FAIL: LLVM :: CodeGen/AArch64/misched-fusion-arith-logic.mir (5654 of
> 29326)
> ******************** TEST 'LLVM ::
> CodeGen/AArch64/misched-fusion-arith-logic.mir' FAILED
> ********************
> shell parser error on: ": 'RUN: at line 1';
> C:\\b\\slave\\clang-x64-windows-msvc\\build\\build\\stage1\\bin\\llc.EXE
> -o -
> C:\\b\\slave\\clang-x64-windows-msvc\\build\\llvm.src\\test\\CodeGen\\AArch64\\misched-fusion-arith-logic.mir
> -mtriple aarch64-unknown -mattr=fuse-arith-logic
> -run-pass=machine-scheduler -misched-print-dags |&
> C:\\b\\slave\\clang-x64-windows-msvc\\build\\build\\stage1\\bin\\FileCheck.EXE
> C:\\b\\slave\\clang-x64-windows-msvc\\build\\llvm.src\\test\\CodeGen\\AArch64\\misched-fusion-arith-logic.mir"
> ********************
>
> I'll submit a quick fix.
>
> On Mon, Jan 14, 2019 at 3:58 PM Evandro Menezes via llvm-commits
> <llvm-commits at lists.llvm.org <mailto:llvm-commits at lists.llvm.org>> wrote:
>
> Author: evandro
> Date: Mon Jan 14 15:54:36 2019
> New Revision: 351139
>
> URL: http://llvm.org/viewvc/llvm-project?rev=351139&view=rev
> Log:
> [AArch64] Add new target feature to fuse arithmetic and logic
> operations
>
> This feature enables the fusion of some arithmetic and logic
> instructions
> together.
>
> Differential revision: https://reviews.llvm.org/D56572
>
> Added:
> llvm/trunk/test/CodeGen/AArch64/misched-fusion-arith-logic.mir
> Modified:
> llvm/trunk/lib/Target/AArch64/AArch64.td
> llvm/trunk/lib/Target/AArch64/AArch64MacroFusion.cpp
> llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h
>
> Modified: llvm/trunk/lib/Target/AArch64/AArch64.td
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AArch64/AArch64.td?rev=351139&r1=351138&r2=351139&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/AArch64/AArch64.td (original)
> +++ llvm/trunk/lib/Target/AArch64/AArch64.td Mon Jan 14 15:54:36 2019
> @@ -188,14 +188,18 @@ def FeatureFuseAES : SubtargetFeature<
> "fuse-aes", "HasFuseAES", "true",
> "CPU fuses AES crypto operations">;
>
> -def FeatureFuseCryptoEOR : SubtargetFeature<
> - "fuse-crypto-eor", "HasFuseCryptoEOR", "true",
> - "CPU fuses AES/PMULL and EOR operations">;
> +def FeatureFuseArithmeticLogic : SubtargetFeature<
> + "fuse-arith-logic", "HasFuseArithmeticLogic", "true",
> + "CPU fuses arithmetic and logic operations">;
>
> def FeatureFuseCCSelect : SubtargetFeature<
> "fuse-csel", "HasFuseCCSelect", "true",
> "CPU fuses conditional select operations">;
>
> +def FeatureFuseCryptoEOR : SubtargetFeature<
> + "fuse-crypto-eor", "HasFuseCryptoEOR", "true",
> + "CPU fuses AES/PMULL and EOR operations">;
> +
> def FeatureFuseLiterals : SubtargetFeature<
> "fuse-literals", "HasFuseLiterals", "true",
> "CPU fuses literal generation operations">;
>
> Modified: llvm/trunk/lib/Target/AArch64/AArch64MacroFusion.cpp
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AArch64/AArch64MacroFusion.cpp?rev=351139&r1=351138&r2=351139&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/AArch64/AArch64MacroFusion.cpp (original)
> +++ llvm/trunk/lib/Target/AArch64/AArch64MacroFusion.cpp Mon Jan
> 14 15:54:36 2019
> @@ -270,7 +270,107 @@ static bool isCCSelectPair(const Machine
> return false;
> }
>
> -/// Check if the instr pair, FirstMI and SecondMI, should be fused
> +// Arithmetic and logic.
> +static bool isArithmeticLogicPair(const MachineInstr *FirstMI,
> + const MachineInstr &SecondMI) {
> + if (AArch64InstrInfo::hasShiftedReg(SecondMI))
> + return false;
> +
> + switch (SecondMI.getOpcode()) {
> + // Arithmetic
> + case AArch64::ADDWrr:
> + case AArch64::ADDXrr:
> + case AArch64::SUBWrr:
> + case AArch64::SUBXrr:
> + case AArch64::ADDWrs:
> + case AArch64::ADDXrs:
> + case AArch64::SUBWrs:
> + case AArch64::SUBXrs:
> + // Logic
> + case AArch64::ANDWrr:
> + case AArch64::ANDXrr:
> + case AArch64::BICWrr:
> + case AArch64::BICXrr:
> + case AArch64::EONWrr:
> + case AArch64::EONXrr:
> + case AArch64::EORWrr:
> + case AArch64::EORXrr:
> + case AArch64::ORNWrr:
> + case AArch64::ORNXrr:
> + case AArch64::ORRWrr:
> + case AArch64::ORRXrr:
> + case AArch64::ANDWrs:
> + case AArch64::ANDXrs:
> + case AArch64::BICWrs:
> + case AArch64::BICXrs:
> + case AArch64::EONWrs:
> + case AArch64::EONXrs:
> + case AArch64::EORWrs:
> + case AArch64::EORXrs:
> + case AArch64::ORNWrs:
> + case AArch64::ORNXrs:
> + case AArch64::ORRWrs:
> + case AArch64::ORRXrs:
> + // Assume the 1st instr to be a wildcard if it is unspecified.
> + if (FirstMI == nullptr)
> + return true;
> +
> + // Arithmetic
> + switch (FirstMI->getOpcode()) {
> + case AArch64::ADDWrr:
> + case AArch64::ADDXrr:
> + case AArch64::ADDSWrr:
> + case AArch64::ADDSXrr:
> + case AArch64::SUBWrr:
> + case AArch64::SUBXrr:
> + case AArch64::SUBSWrr:
> + case AArch64::SUBSXrr:
> + return true;
> + case AArch64::ADDWrs:
> + case AArch64::ADDXrs:
> + case AArch64::ADDSWrs:
> + case AArch64::ADDSXrs:
> + case AArch64::SUBWrs:
> + case AArch64::SUBXrs:
> + case AArch64::SUBSWrs:
> + case AArch64::SUBSXrs:
> + return !AArch64InstrInfo::hasShiftedReg(*FirstMI);
> + }
> + break;
> +
> + // Arithmetic, setting flags.
> + case AArch64::ADDSWrr:
> + case AArch64::ADDSXrr:
> + case AArch64::SUBSWrr:
> + case AArch64::SUBSXrr:
> + case AArch64::ADDSWrs:
> + case AArch64::ADDSXrs:
> + case AArch64::SUBSWrs:
> + case AArch64::SUBSXrs:
> + // Assume the 1st instr to be a wildcard if it is unspecified.
> + if (FirstMI == nullptr)
> + return true;
> +
> + // Arithmetic, not setting flags.
> + switch (FirstMI->getOpcode()) {
> + case AArch64::ADDWrr:
> + case AArch64::ADDXrr:
> + case AArch64::SUBWrr:
> + case AArch64::SUBXrr:
> + return true;
> + case AArch64::ADDWrs:
> + case AArch64::ADDXrs:
> + case AArch64::SUBWrs:
> + case AArch64::SUBXrs:
> + return !AArch64InstrInfo::hasShiftedReg(*FirstMI);
> + }
> + break;
> + }
> +
> + return false;
> +}
> +
> +/// \brief Check if the instr pair, FirstMI and SecondMI, should
> be fused
> /// together. Given SecondMI, when FirstMI is unspecified, then
> check if
> /// SecondMI may be part of a fused pair at all.
> static bool shouldScheduleAdjacent(const TargetInstrInfo &TII,
> @@ -295,6 +395,8 @@ static bool shouldScheduleAdjacent(const
> return true;
> if (ST.hasFuseCCSelect() && isCCSelectPair(FirstMI, SecondMI))
> return true;
> + if (ST.hasFuseArithmeticLogic() &&
> isArithmeticLogicPair(FirstMI, SecondMI))
> + return true;
>
> return false;
> }
>
> Modified: llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h?rev=351139&r1=351138&r2=351139&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h (original)
> +++ llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h Mon Jan 14
> 15:54:36 2019
> @@ -166,8 +166,9 @@ protected:
> bool HasArithmeticCbzFusion = false;
> bool HasFuseAddress = false;
> bool HasFuseAES = false;
> - bool HasFuseCryptoEOR = false;
> + bool HasFuseArithmeticLogic = false;
> bool HasFuseCCSelect = false;
> + bool HasFuseCryptoEOR = false;
> bool HasFuseLiterals = false;
> bool DisableLatencySchedHeuristic = false;
> bool UseRSqrt = false;
> @@ -311,14 +312,16 @@ public:
> bool hasArithmeticCbzFusion() const { return
> HasArithmeticCbzFusion; }
> bool hasFuseAddress() const { return HasFuseAddress; }
> bool hasFuseAES() const { return HasFuseAES; }
> - bool hasFuseCryptoEOR() const { return HasFuseCryptoEOR; }
> + bool hasFuseArithmeticLogic() const { return
> HasFuseArithmeticLogic; }
> bool hasFuseCCSelect() const { return HasFuseCCSelect; }
> + bool hasFuseCryptoEOR() const { return HasFuseCryptoEOR; }
> bool hasFuseLiterals() const { return HasFuseLiterals; }
>
> /// Return true if the CPU supports any kind of instruction fusion.
> bool hasFusion() const {
> return hasArithmeticBccFusion() || hasArithmeticCbzFusion() ||
> - hasFuseAES() || hasFuseCCSelect() || hasFuseLiterals();
> + hasFuseAES() || hasFuseArithmeticLogic() ||
> + hasFuseCCSelect() || hasFuseLiterals();
> }
>
> bool useRSqrt() const { return UseRSqrt; }
>
> Added: llvm/trunk/test/CodeGen/AArch64/misched-fusion-arith-logic.mir
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AArch64/misched-fusion-arith-logic.mir?rev=351139&view=auto
> ==============================================================================
> --- llvm/trunk/test/CodeGen/AArch64/misched-fusion-arith-logic.mir
> (added)
> +++ llvm/trunk/test/CodeGen/AArch64/misched-fusion-arith-logic.mir
> Mon Jan 14 15:54:36 2019
> @@ -0,0 +1,111 @@
> +# RUN: llc -o - %s -mtriple aarch64-unknown
> -mattr=fuse-arith-logic -run-pass=machine-scheduler
> -misched-print-dags |& FileCheck %s
> +# REQUIRES: asserts
> +
> +---
> +name: arith
> +body: |
> + bb.0.entry:
> + %0:gpr32 = SUBWrr undef $w0, undef $w1
> + %1:gpr32 = ADDWrr undef $w1, undef $w2
> + %2:gpr32 = SUBWrs %0, undef $w2, 0
> + %3:gpr32 = ADDWrs %1, undef $w3, 0
> +
> + ; CHECK: SU(0): %0:gpr32 = SUBWrr undef $w0, undef $w1
> + ; CHECK: Successors:
> + ; CHECK: SU(2): Ord Latency=0 Cluster
> + ; CHECK: SU(1): %1:gpr32 = ADDWrr undef $w1, undef $w2
> + ; CHECK: Successors:
> + ; CHECK: SU(3): Ord Latency=0 Cluster
> + ; CHECK: SU(2): dead %2:gpr32 = SUBWrs %0:gpr32, undef $w2, 0
> + ; CHECK: Predecessors:
> + ; CHECK: SU(0): Ord Latency=0 Cluster
> + ; CHECK: SU(3): dead %3:gpr32 = ADDWrs %1:gpr32, undef $w3, 0
> + ; CHECK: Predecessors:
> + ; CHECK: SU(1): Ord Latency=0 Cluster
> +...
> +---
> +name: compare
> +body: |
> + bb.0.entry:
> + %0:gpr64 = ADDXrr undef $x0, undef $x1
> + %1:gpr64 = SUBXrs undef $x1, undef $x2, 0
> + %2:gpr64 = ADDSXrr %0, undef $x3, implicit-def $nzcv
> + %3:gpr64 = SUBSXrs %1, undef $x4, 0, implicit-def $nzcv
> +
> + ; CHECK: SU(0): %0:gpr64 = ADDXrr undef $x0, undef $x1
> + ; CHECK: Successors:
> + ; CHECK: SU(2): Ord Latency=0 Cluster
> + ; CHECK: SU(1): %1:gpr64 = SUBXrs undef $x1, undef $x2, 0
> + ; CHECK: Successors:
> + ; CHECK: SU(3): Ord Latency=0 Cluster
> + ; CHECK: SU(2): dead %2:gpr64 = ADDSXrr %0:gpr64, undef $x3,
> implicit-def $nzcv
> + ; CHECK: Predecessors:
> + ; CHECK: SU(0): Ord Latency=0 Cluster
> + ; CHECK: SU(3): dead %3:gpr64 = SUBSXrs %1:gpr64, undef $x4,
> 0, implicit-def $nzcv
> + ; CHECK: Predecessors:
> + ; CHECK: SU(1): Ord Latency=0 Cluster
> +...
> +---
> +name: logic
> +body: |
> + bb.0.entry:
> + %0:gpr32 = ADDWrr undef $w0, undef $w1
> + %1:gpr64 = SUBXrs undef $x1, undef $x2, 0
> + %3:gpr32 = ANDWrs %0, undef $w3, 0
> + %4:gpr64 = ORRXrr %1, undef $x4
> +
> + ; CHECK: SU(0): %0:gpr32 = ADDWrr undef $w0, undef $w1
> + ; CHECK: Successors:
> + ; CHECK: SU(2): Ord Latency=0 Cluster
> + ; CHECK: SU(1): %1:gpr64 = SUBXrs undef $x1, undef $x2, 0
> + ; CHECK: Successors:
> + ; CHECK: SU(3): Ord Latency=0 Cluster
> + ; CHECK: SU(2): dead %2:gpr32 = ANDWrs %0:gpr32, undef $w3, 0
> + ; CHECK: Predecessors:
> + ; CHECK: SU(0): Ord Latency=0 Cluster
> + ; CHECK: SU(3): dead %3:gpr64 = ORRXrr %1:gpr64, undef $x4
> + ; CHECK: Predecessors:
> + ; CHECK: SU(1): Ord Latency=0 Cluster
> +...
> +---
> +name: nope
> +body: |
> + bb.0.entry:
> + ; Shifted register.
> + %0:gpr32 = SUBWrr undef $w0, undef $w1
> + %1:gpr32 = SUBWrs %0, undef $w2, 1
> + ; CHECKSU(0)%0:gpr32 = SUBWrr undef $w0, undef $w1
> + ; CHECKSuccessors:
> + ; CHECK-NOTSU(1)Ord Latency=0 Cluster
> + ; CHECKSU(1)dead %1:gpr32 = SUBWrs %0:gpr32, undef $w2, 1
> +
> + ; Multiple successors.
> + %2:gpr64 = ADDXrr undef $x0, undef $x1
> + %3:gpr32 = EXTRACT_SUBREG %2, %subreg.sub_32
> + %4:gpr32 = ANDWrs %3, undef $w2, 0
> + %5:gpr64 = ADDSXrr %2, undef $x3, implicit-def $nzcv
> + ; CHECKSU(2)%2:gpr64 = ADDXrr undef $x0, undef $x1
> + ; CHECKSuccessors:
> + ; CHECK-NOTSU(3)Ord Latency=0 Cluster
> + ; CHECKSU(5)Ord Latency=0 Cluster
> + ; CHECKSU(3)%3:gpr32 = EXTRACT_SUBREG %2:gpr64, %subreg.sub_32
> + ; CHECKSU(5)dead %5:gpr64 = ADDSXrr %2:gpr64, undef $x3,
> implicit-def $nzcv
> +
> + ; Different register sizes.
> + %6:gpr32 = SUBWrr undef $w0, undef $w1
> + %7:gpr64 = ADDXrr undef $x1, undef $x2
> + %8:gpr64 = SUBXrr %7, undef $x3
> + %9:gpr32 = ADDWrr %6, undef $w4
> + ; CHECKSU(6)%6:gpr32 = SUBWrr undef $w0, undef $w1
> + ; CHECKSuccessors:
> + ; CHECK-NOTSU(8)Ord Latency=0 Cluster
> + ; CHECKSU(7)%7:gpr64 = ADDXrr undef $x1, undef $x2
> + ; CHECKSuccessors:
> + ; CHECK-NOTSU(9)Ord Latency=0 Cluster
> + ; CHECKSU(8)dead %8:gpr64 = SUBXrr %7:gpr64, undef $x3
> + ; CHECKPredecessors:
> + ; CHECKSU(7)Ord Latency=0 Cluster
> + ; CHECKSU(9)dead %9:gpr32 = ADDWrr %6:gpr32, undef $w4
> + ; CHECKPredecessors:
> + ; CHECKSU(6)Ord Latency=0 Cluster
> +...
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org <mailto:llvm-commits at lists.llvm.org>
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190114/f219fda1/attachment-0001.html>
More information about the llvm-commits
mailing list