f27017a063
The LA464 micro-architecture is very sensitive to alignment of hot code, with performance variation of up to ~12% in the go1 benchmark suite of the Go language (as observed by me during my work on the Go loong64 port). [[ https://go.dev/cl/479816 | Manual alignment of certain loops ]] and [[ https://go.dev/cl/479817 | automatic alignment of loop heads ]] helps a lot there, by reducing much of the random variation and generally increasing performance, so we naturally want to do the same here. Practically, LA464 is the only LoongArch micro-architecture in wide use, and we are currently supporting just that. The first "4" in "LA464" stands for "4-issue", in particular its instruction fetch and decode stages are 4-wide; so functions and branch targets should be preferably aligned to at least 16 bytes for best throughput. The Loongson team has benchmarked various combinations of function, loop, and branch target alignments with GCC. [[ https://gcc.gnu.org/pipermail/gcc-patches/2023-May/619980.html | The results ]] show that "16-byte label alignment together with 32-byte function alignment gives best results in terms of SPEC score". A "label" in GCC means a branch target; while we don't currently align branch targets, we do align loops, so in this patch we default to 32-byte function alignment and 16-byte loop alignment. Reviewed By: SixWeining Differential Revision: https://reviews.llvm.org/D148622
92 lines
3.3 KiB
C++
92 lines
3.3 KiB
C++
//===-- LoongArchSubtarget.cpp - LoongArch Subtarget Information -*- C++ -*--=//
|
|
//
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This file implements the LoongArch specific subclass of TargetSubtargetInfo.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include "LoongArchSubtarget.h"
|
|
#include "LoongArchFrameLowering.h"
|
|
#include "MCTargetDesc/LoongArchBaseInfo.h"
|
|
|
|
using namespace llvm;
|
|
|
|
#define DEBUG_TYPE "loongarch-subtarget"
|
|
|
|
#define GET_SUBTARGETINFO_TARGET_DESC
|
|
#define GET_SUBTARGETINFO_CTOR
|
|
#include "LoongArchGenSubtargetInfo.inc"
|
|
|
|
void LoongArchSubtarget::anchor() {}
|
|
|
|
LoongArchSubtarget &LoongArchSubtarget::initializeSubtargetDependencies(
|
|
const Triple &TT, StringRef CPU, StringRef TuneCPU, StringRef FS,
|
|
StringRef ABIName) {
|
|
bool Is64Bit = TT.isArch64Bit();
|
|
if (CPU.empty() || CPU == "generic")
|
|
CPU = Is64Bit ? "generic-la64" : "generic-la32";
|
|
|
|
if (TuneCPU.empty())
|
|
TuneCPU = CPU;
|
|
|
|
ParseSubtargetFeatures(CPU, TuneCPU, FS);
|
|
initializeProperties(TuneCPU);
|
|
if (Is64Bit) {
|
|
GRLenVT = MVT::i64;
|
|
GRLen = 64;
|
|
}
|
|
|
|
if (HasLA32 == HasLA64)
|
|
report_fatal_error("Please use one feature of 32bit and 64bit.");
|
|
|
|
if (Is64Bit && HasLA32)
|
|
report_fatal_error("Feature 32bit should be used for loongarch32 target.");
|
|
|
|
if (!Is64Bit && HasLA64)
|
|
report_fatal_error("Feature 64bit should be used for loongarch64 target.");
|
|
|
|
TargetABI = LoongArchABI::computeTargetABI(TT, ABIName);
|
|
|
|
return *this;
|
|
}
|
|
|
|
void LoongArchSubtarget::initializeProperties(StringRef TuneCPU) {
|
|
// Initialize CPU specific properties. We should add a tablegen feature for
|
|
// this in the future so we can specify it together with the subtarget
|
|
// features.
|
|
|
|
// TODO: Check TuneCPU and override defaults (that are for LA464) once we
|
|
// support optimizing for more uarchs.
|
|
|
|
// Default to the alignment settings empirically confirmed to perform best
|
|
// on LA464, with 4-wide instruction fetch and decode stages. These settings
|
|
// can also be overridden in initializeProperties.
|
|
//
|
|
// We default to such higher-than-minimum alignments because we assume that:
|
|
//
|
|
// * these settings should benefit most existing uarchs/users,
|
|
// * future general-purpose LoongArch cores are likely to have issue widths
|
|
// equal to or wider than 4,
|
|
// * instruction sequences best for LA464 should not pessimize other future
|
|
// uarchs, and
|
|
// * narrower cores would not suffer much (aside from slightly increased
|
|
// ICache footprint maybe), compared to the gains everywhere else.
|
|
PrefFunctionAlignment = Align(32);
|
|
PrefLoopAlignment = Align(16);
|
|
MaxBytesForAlignment = 16;
|
|
}
|
|
|
|
LoongArchSubtarget::LoongArchSubtarget(const Triple &TT, StringRef CPU,
|
|
StringRef TuneCPU, StringRef FS,
|
|
StringRef ABIName,
|
|
const TargetMachine &TM)
|
|
: LoongArchGenSubtargetInfo(TT, CPU, TuneCPU, FS),
|
|
FrameLowering(
|
|
initializeSubtargetDependencies(TT, CPU, TuneCPU, FS, ABIName)),
|
|
InstrInfo(*this), RegInfo(getHwMode()), TLInfo(TM, *this) {}
|